{"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 3 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have three packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 65,954,164 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that sixty-five million nine hundred fifty-four thousand one hundred sixty-four people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 819,521 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that eight hundred nineteen thousand five hundred twenty-one people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 279,190 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that two hundred seventy-nine thousand one hundred ninety people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 45,729 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that forty-five thousand seven hundred twenty-nine people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 235 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have two hundred thirty-five packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 11.50AM on 20-04-2013, room 017839."}, {"role": "assistant", "content": "The meeting is at eleven fifty A M on the twentieth of April twenty thirteen, room zero one seven eight three nine."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 216 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have two hundred sixteen packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 640 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have six hundred forty packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 72,146,213 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that seventy-two million one hundred forty-six thousand two hundred thirteen people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 476,520 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that four hundred seventy-six thousand five hundred twenty people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 62,321 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that sixty-two thousand three hundred twenty-one people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 11th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the eleventh stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at dilani@outlook.sa."}, {"role": "assistant", "content": "For further assistance, please email us at dilani at outlook dot sa."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at maria@sapo.pt."}, {"role": "assistant", "content": "For further assistance, please email us at maria at sapo dot pt."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at dilani@company.co."}, {"role": "assistant", "content": "For further assistance, please email us at dilani at company dot co."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at ali@sapo.pt."}, {"role": "assistant", "content": "For further assistance, please email us at ali at sapo dot pt."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 8th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the eighth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 47 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have forty-seven packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 5,833,502 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that five million eight hundred thirty-three thousand five hundred two people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 6 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have six packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at jean.dupont@gmail.co.id."}, {"role": "assistant", "content": "For further assistance, please email us at jean dot dupont at gmail dot co dot id."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 8611 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that eight thousand six hundred eleven people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 39,284,971 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that thirty-nine million two hundred eighty-four thousand nine hundred seventy-one people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 50,592 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that fifty thousand five hundred ninety-two people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 5th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the fifth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at luca@outlook.sa."}, {"role": "assistant", "content": "For further assistance, please email us at luca at outlook dot sa."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kumar@outlook.sa."}, {"role": "assistant", "content": "For further assistance, please email us at kumar at outlook dot sa."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 5.05PM on 23 March 2016, room 215847."}, {"role": "assistant", "content": "The meeting is at five zero five P M on the twenty-third of March twenty sixteen, room two one five eight four seven."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 6713 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that six thousand seven hundred thirteen people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 10a.m. on 22-10-2032, room 1992."}, {"role": "assistant", "content": "The meeting is at ten a m on the twenty-second of October twenty thirty-two, room nineteen ninety-two."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 23:45 on 3/12/2010, room 919015."}, {"role": "assistant", "content": "The meeting is at twenty-three forty-five on the third of December twenty ten, room nine one nine zero one five."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 30th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the thirtieth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 16th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the sixteenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 3429 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that three thousand four hundred twenty-nine people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kasia@example.com."}, {"role": "assistant", "content": "For further assistance, please email us at kasia at example dot com."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at ravi@example.com."}, {"role": "assistant", "content": "For further assistance, please email us at ravi at example dot com."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 1,327,168 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that one million three hundred twenty-seven thousand one hundred sixty-eight people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 7th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the seventh stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 14th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the fourteenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 7,220 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that seven thousand two hundred twenty people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 9,022 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that nine thousand twenty-two people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 35,926,996 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that thirty-five million nine hundred twenty-six thousand nine hundred ninety-six people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 15th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the fifteenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 12th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twelfth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 297,991 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that two hundred ninety-seven thousand nine hundred ninety-one people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 3,415 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that three thousand four hundred fifteen people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at billing@onet.pl."}, {"role": "assistant", "content": "For further assistance, please email us at billing at onet dot pl."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at ahmad.b@ziggo.nl."}, {"role": "assistant", "content": "For further assistance, please email us at ahmad dot b at ziggo dot nl."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 78 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have seventy-eight packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 924,013 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that nine hundred twenty-four thousand thirteen people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 396 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have three hundred ninety-six packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at pedro@correo.es."}, {"role": "assistant", "content": "For further assistance, please email us at pedro at correo dot es."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 5p.m. on 2 January 2016, room 8576."}, {"role": "assistant", "content": "The meeting is at five p m on the second of January twenty sixteen, room eight five seven six."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 834,345 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that eight hundred thirty-four thousand three hundred forty-five people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 7:45 PM on 5 September, room 982588."}, {"role": "assistant", "content": "The meeting is at seven forty-five P M on the fifth of September, room nine eight two five eight eight."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at sofia@correo.es."}, {"role": "assistant", "content": "For further assistance, please email us at sofia at correo dot es."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 5:00 on 6/10/2001, room 0943."}, {"role": "assistant", "content": "The meeting is at five o'clock on the sixth of October two thousand and one, room zero nine four three."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kasia@gmail.co.id."}, {"role": "assistant", "content": "For further assistance, please email us at kasia at gmail dot co dot id."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at info@example.com."}, {"role": "assistant", "content": "For further assistance, please email us at info at example dot com."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 12.00am on 4/1/2012, room 1313."}, {"role": "assistant", "content": "The meeting is at twelve a m on the fourth of January twenty twelve, room one three one three."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kumar@yahoo.com.ph."}, {"role": "assistant", "content": "For further assistance, please email us at kumar at yahoo dot com dot ph."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 4th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the fourth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 869 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have eight hundred sixty-nine packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 33,077 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that thirty-three thousand seventy-seven people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 64,525 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that sixty-four thousand five hundred twenty-five people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 90,774,142 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that ninety million seven hundred seventy-four thousand one hundred forty-two people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 10th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the tenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 349 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have three hundred forty-nine packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at info66@gmail.co.id."}, {"role": "assistant", "content": "For further assistance, please email us at info six six at gmail dot co dot id."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 4868 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that four thousand eight hundred sixty-eight people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 11:10 PM on 3 April, room 24226."}, {"role": "assistant", "content": "The meeting is at eleven ten P M on the third of April, room two four two two six."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 76,481 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that seventy-six thousand four hundred eighty-one people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 49,489 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that forty-nine thousand four hundred eighty-nine people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at nurul_h@sltnet.lk."}, {"role": "assistant", "content": "For further assistance, please email us at nurul underscore h at sltnet dot lk."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at omar@onet.pl."}, {"role": "assistant", "content": "For further assistance, please email us at omar at onet dot pl."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 18th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the eighteenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 81 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have eighty-one packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 3pm on 17/3/2013, room 5529."}, {"role": "assistant", "content": "The meeting is at three p m on the seventeenth of March twenty thirteen, room five five two nine."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 32,129 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that thirty-two thousand one hundred twenty-nine people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 26th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twenty-sixth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 97,225 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that ninety-seven thousand two hundred twenty-five people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at sofia@sltnet.lk."}, {"role": "assistant", "content": "For further assistance, please email us at sofia at sltnet dot lk."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 31st stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the thirty-first stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 4:05 a.m. on 14 December 2017, room 8468."}, {"role": "assistant", "content": "The meeting is at four zero five a m on the fourteenth of December twenty seventeen, room eight four six eight."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 21st stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twenty-first stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 908 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have nine hundred eight packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 14 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have fourteen packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 20th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twentieth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 170,957 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that one hundred seventy thousand nine hundred fifty-seven people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 35 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have thirty-five packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 34,065,276 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that thirty-four million sixty-five thousand two hundred seventy-six people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at anna@example.com."}, {"role": "assistant", "content": "For further assistance, please email us at anna at example dot com."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 937 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have nine hundred thirty-seven packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kasia@yahoo.com."}, {"role": "assistant", "content": "For further assistance, please email us at kasia at yahoo dot com."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 12:40 pm on April 27, 2012, room 99760."}, {"role": "assistant", "content": "The meeting is at twelve forty p m on April twenty-seventh, twenty twelve, room nine nine seven six zero."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at billing@mail.fr."}, {"role": "assistant", "content": "For further assistance, please email us at billing at mail dot fr."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at kasia@ziggo.nl."}, {"role": "assistant", "content": "For further assistance, please email us at kasia at ziggo dot nl."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 72 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have seventy-two packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 4:15 AM on 11/5/2028, room 6220."}, {"role": "assistant", "content": "The meeting is at four fifteen A M on the eleventh of May twenty twenty-eight, room six two two zero."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 11:15 pm on January 5, 2028, room 04891."}, {"role": "assistant", "content": "The meeting is at eleven fifteen p m on January fifth, twenty twenty-eight, room zero four eight nine one."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 28,434,329 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that twenty-eight million four hundred thirty-four thousand three hundred twenty-nine people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 23rd stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twenty-third stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 27th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the twenty-seventh stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 64 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have sixty-four packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 2 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have two packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 236 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have two hundred thirty-six packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 787 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have seven hundred eighty-seven packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "You have reached the 13th stage of your treatment plan."}, {"role": "assistant", "content": "You have reached the thirteenth stage of your treatment plan."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at omar@correo.es."}, {"role": "assistant", "content": "For further assistance, please email us at omar at correo dot es."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 55,747,220 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that fifty-five million seven hundred forty-seven thousand two hundred twenty people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at maria@outlook.sa."}, {"role": "assistant", "content": "For further assistance, please email us at maria at outlook dot sa."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 859 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have eight hundred fifty-nine packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 8:10 AM on 3/9/2021, room 754000."}, {"role": "assistant", "content": "The meeting is at eight ten A M on the third of September twenty twenty-one, room seven five four zero zero zero."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 4 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have four packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "For further assistance, please email us at info8@sapo.pt."}, {"role": "assistant", "content": "For further assistance, please email us at info eight at sapo dot pt."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 881,242 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that eight hundred eighty-one thousand two hundred forty-two people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The meeting is at 6a.m. on 22 April, room 2642."}, {"role": "assistant", "content": "The meeting is at six a m on the twenty-second of April, room two six four two."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 74 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have seventy-four packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Our records show that 4,484 people have already registered for the new program."}, {"role": "assistant", "content": "Our records show that four thousand four hundred eighty-four people have already registered for the new program."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 5 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have five packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 819 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have eight hundred nineteen packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We have 7 packages waiting for you at the pickup point."}, {"role": "assistant", "content": "We have seven packages waiting for you at the pickup point."}], "lang": "en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 37. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tiga puluh tujuh ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 437232. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah empat tiga tujuh dua tiga dua. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 91.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal sembilan puluh satu ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 76929. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tujuh enam sembilan dua sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan ABC1234 anda sedang diproses dan akan dihantar dalam masa 9 hari bekerja."}, {"role": "assistant", "content": "Pesanan ABC satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa sembilan hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan X-12340567 anda sedang diproses dan akan dihantar dalam masa 981 hari bekerja."}, {"role": "assistant", "content": "Pesanan X satu dua tiga empat kosong lima enam tujuh anda sedang diproses dan akan dihantar dalam masa sembilan ratus lapan puluh satu hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 8.4 Mbps, iaitu peningkatan sebanyak 88% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lapan perpuluhan empat megabit sesaat, iaitu peningkatan sebanyak lapan puluh lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan A12 anda sedang diproses dan akan dihantar dalam masa 63 hari bekerja."}, {"role": "assistant", "content": "Pesanan A dua belas anda sedang diproses dan akan dihantar dalam masa enam puluh tiga hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 77451. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tujuh tujuh empat lima satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM17. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh belas ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 39.18 Mbps, iaitu peningkatan sebanyak 89% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah tiga puluh sembilan perpuluhan satu lapan megabit sesaat, iaitu peningkatan sebanyak lapan puluh sembilan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 68.99 Mbps, iaitu peningkatan sebanyak 21.8% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah enam puluh lapan perpuluhan sembilan sembilan megabit sesaat, iaitu peningkatan sebanyak dua puluh satu perpuluhan lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan REF-2024-0113 anda sedang diproses dan akan dihantar dalam masa 1 hari bekerja."}, {"role": "assistant", "content": "Pesanan REF dua kosong dua empat kosong satu satu tiga anda sedang diproses dan akan dihantar dalam masa satu hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 39475. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tiga sembilan empat tujuh lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 2,426. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal dua ribu empat ratus dua puluh enam ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 30,296. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tiga puluh ribu dua ratus sembilan puluh enam ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 5648. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lima enam empat lapan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 84341. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lapan empat tiga empat satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 59.41 Mbps, iaitu peningkatan sebanyak 12% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lima puluh sembilan perpuluhan empat satu megabit sesaat, iaitu peningkatan sebanyak dua belas peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 98226. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah sembilan lapan dua dua enam. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 44.82 Mbps, iaitu peningkatan sebanyak 27% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah empat puluh empat perpuluhan lapan dua megabit sesaat, iaitu peningkatan sebanyak dua puluh tujuh peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 30530. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah tiga kosong lima tiga kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 541940. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lima empat satu sembilan empat kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 84216. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lapan empat dua satu enam. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan ABC1234 anda sedang diproses dan akan dihantar dalam masa 830 hari bekerja."}, {"role": "assistant", "content": "Pesanan ABC satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa lapan ratus tiga puluh hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 95. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal sembilan puluh lima ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 526649. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lima dua enam enam empat sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 53922. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lima tiga sembilan dua dua. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan WXY 1234 anda sedang diproses dan akan dihantar dalam masa 18 hari bekerja."}, {"role": "assistant", "content": "Pesanan WXY satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa lapan belas hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 877124. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lapan tujuh tujuh satu dua empat. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 102677. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah satu kosong dua enam tujuh tujuh. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 93.96 Mbps, iaitu peningkatan sebanyak 3.7% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan puluh tiga perpuluhan sembilan enam megabit sesaat, iaitu peningkatan sebanyak tiga perpuluhan tujuh peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 15,586. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal lima belas ribu lima ratus lapan puluh enam ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan B2-14 anda sedang diproses dan akan dihantar dalam masa 5 hari bekerja."}, {"role": "assistant", "content": "Pesanan B dua empat belas anda sedang diproses dan akan dihantar dalam masa lima hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 66.44 Mbps, iaitu peningkatan sebanyak 62% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah enam puluh enam perpuluhan empat empat megabit sesaat, iaitu peningkatan sebanyak enam puluh dua peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 96967. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah sembilan enam sembilan enam tujuh. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM31,223.98. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tiga puluh satu ribu dua ratus dua puluh tiga ringgit sembilan puluh lapan sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM6.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal enam ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan WXY 1234 anda sedang diproses dan akan dihantar dalam masa 3 hari bekerja."}, {"role": "assistant", "content": "Pesanan WXY satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa tiga hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 1803. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah seribu lapan ratus tiga. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 73.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh puluh tiga ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan PO 55219 anda sedang diproses dan akan dihantar dalam masa 17 hari bekerja."}, {"role": "assistant", "content": "Pesanan PO lima lima dua satu sembilan anda sedang diproses dan akan dihantar dalam masa tujuh belas hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan QZ8501 anda sedang diproses dan akan dihantar dalam masa 896 hari bekerja."}, {"role": "assistant", "content": "Pesanan QZ lapan lima kosong satu anda sedang diproses dan akan dihantar dalam masa lapan ratus sembilan puluh enam hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 965621. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah sembilan enam lima enam dua satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 63.79 Mbps, iaitu peningkatan sebanyak 11% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah enam puluh tiga perpuluhan tujuh sembilan megabit sesaat, iaitu peningkatan sebanyak sebelas peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 79,677.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh puluh sembilan ribu enam ratus tujuh puluh tujuh ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 10.5 Mbps, iaitu peningkatan sebanyak 43% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sepuluh perpuluhan lima megabit sesaat, iaitu peningkatan sebanyak empat puluh tiga peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan AK6120 anda sedang diproses dan akan dihantar dalam masa 164 hari bekerja."}, {"role": "assistant", "content": "Pesanan AK enam satu dua kosong anda sedang diproses dan akan dihantar dalam masa seratus enam puluh empat hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 3697. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tiga enam sembilan tujuh. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan AK6120 anda sedang diproses dan akan dihantar dalam masa 19 hari bekerja."}, {"role": "assistant", "content": "Pesanan AK enam satu dua kosong anda sedang diproses dan akan dihantar dalam masa sembilan belas hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 74.95 Mbps, iaitu peningkatan sebanyak 24% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah tujuh puluh empat perpuluhan sembilan lima megabit sesaat, iaitu peningkatan sebanyak dua puluh empat peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 183295. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah satu lapan tiga dua sembilan lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 51.09 Mbps, iaitu peningkatan sebanyak 1.9% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lima puluh satu perpuluhan kosong sembilan megabit sesaat, iaitu peningkatan sebanyak satu perpuluhan sembilan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 95.64 Mbps, iaitu peningkatan sebanyak 49.7% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan puluh lima perpuluhan enam empat megabit sesaat, iaitu peningkatan sebanyak empat puluh sembilan perpuluhan tujuh peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 26.5 Mbps, iaitu peningkatan sebanyak 13.8% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah dua puluh enam perpuluhan lima megabit sesaat, iaitu peningkatan sebanyak tiga belas perpuluhan lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM10,022.50. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal sepuluh ribu dua puluh dua ringgit lima puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 45311. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah empat lima tiga satu satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 18215. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah satu lapan dua satu lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan X-12340567 anda sedang diproses dan akan dihantar dalam masa 164 hari bekerja."}, {"role": "assistant", "content": "Pesanan X satu dua tiga empat kosong lima enam tujuh anda sedang diproses dan akan dihantar dalam masa seratus enam puluh empat hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan TK-88213 anda sedang diproses dan akan dihantar dalam masa 13 hari bekerja."}, {"role": "assistant", "content": "Pesanan TK lapan lapan dua satu tiga anda sedang diproses dan akan dihantar dalam masa tiga belas hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 73.46 Mbps, iaitu peningkatan sebanyak 88% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah tujuh puluh tiga perpuluhan empat enam megabit sesaat, iaitu peningkatan sebanyak lapan puluh lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 50549. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lima kosong lima empat sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 83.19 Mbps, iaitu peningkatan sebanyak 47.5% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lapan puluh tiga perpuluhan satu sembilan megabit sesaat, iaitu peningkatan sebanyak empat puluh tujuh perpuluhan lima peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 98,377. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal sembilan puluh lapan ribu tiga ratus tujuh puluh tujuh ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 23.62 Mbps, iaitu peningkatan sebanyak 98.9% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah dua puluh tiga perpuluhan enam dua megabit sesaat, iaitu peningkatan sebanyak sembilan puluh lapan perpuluhan sembilan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 6104. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah enam satu kosong empat. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM949.50. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal sembilan ratus empat puluh sembilan ringgit lima puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 22745. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah dua dua tujuh empat lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 55.77 Mbps, iaitu peningkatan sebanyak 3.5% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lima puluh lima perpuluhan tujuh tujuh megabit sesaat, iaitu peningkatan sebanyak tiga perpuluhan lima peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 89446. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah lapan sembilan empat empat enam. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 97.08 Mbps, iaitu peningkatan sebanyak 50.4% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan puluh tujuh perpuluhan kosong lapan megabit sesaat, iaitu peningkatan sebanyak lima puluh perpuluhan empat peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM410.50. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal empat ratus sepuluh ringgit lima puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 048369. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah kosong empat lapan tiga enam sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan WXY 1234 anda sedang diproses dan akan dihantar dalam masa 43 hari bekerja."}, {"role": "assistant", "content": "Pesanan WXY satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa empat puluh tiga hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 9526. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah sembilan lima dua enam. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 2490. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah dua empat sembilan kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 77666. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tujuh tujuh enam enam enam. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 076752. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah kosong tujuh enam tujuh lima dua. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 66280. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah enam enam dua lapan kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 095985. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah kosong sembilan lima sembilan lapan lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan AK6120 anda sedang diproses dan akan dihantar dalam masa 2 hari bekerja."}, {"role": "assistant", "content": "Pesanan AK enam satu dua kosong anda sedang diproses dan akan dihantar dalam masa dua hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 176.71. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal seratus tujuh puluh enam ringgit tujuh puluh satu sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 77.2 Mbps, iaitu peningkatan sebanyak 2% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah tujuh puluh tujuh perpuluhan dua megabit sesaat, iaitu peningkatan sebanyak dua peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM32,024. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tiga puluh dua ribu dua puluh empat ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 92.83 Mbps, iaitu peningkatan sebanyak 78% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan puluh dua perpuluhan lapan tiga megabit sesaat, iaitu peningkatan sebanyak tujuh puluh lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 90611. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah sembilan kosong enam satu satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 60520. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah enam kosong lima dua kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan WXY 1234 anda sedang diproses dan akan dihantar dalam masa 75 hari bekerja."}, {"role": "assistant", "content": "Pesanan WXY satu dua tiga empat anda sedang diproses dan akan dihantar dalam masa tujuh puluh lima hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM138. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal seratus tiga puluh lapan ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 3215. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah tiga dua satu lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 850737. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lapan lima kosong tujuh tiga tujuh. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM56. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal lima puluh enam ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 349448. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah tiga empat sembilan empat empat lapan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM76.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh puluh enam ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 645517. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah enam empat lima lima satu tujuh. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan INV0472 anda sedang diproses dan akan dihantar dalam masa 495 hari bekerja."}, {"role": "assistant", "content": "Pesanan INV kosong empat tujuh dua anda sedang diproses dan akan dihantar dalam masa empat ratus sembilan puluh lima hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM28,044.90. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal dua puluh lapan ribu empat puluh empat ringgit sembilan puluh sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 50.72 Mbps, iaitu peningkatan sebanyak 11% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lima puluh perpuluhan tujuh dua megabit sesaat, iaitu peningkatan sebanyak sebelas peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 19.33 Mbps, iaitu peningkatan sebanyak 75.9% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan belas perpuluhan tiga tiga megabit sesaat, iaitu peningkatan sebanyak tujuh puluh lima perpuluhan sembilan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 39,516.41. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tiga puluh sembilan ribu lima ratus enam belas ringgit empat puluh satu sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 3355. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah tiga tiga lima lima. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 78. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh puluh lapan ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 87.79 Mbps, iaitu peningkatan sebanyak 37% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lapan puluh tujuh perpuluhan tujuh sembilan megabit sesaat, iaitu peningkatan sebanyak tiga puluh tujuh peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 58370. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lima lapan tiga tujuh kosong. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan MH370 anda sedang diproses dan akan dihantar dalam masa 1 hari bekerja."}, {"role": "assistant", "content": "Pesanan MH tiga tujuh kosong anda sedang diproses dan akan dihantar dalam masa satu hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 10.6 Mbps, iaitu peningkatan sebanyak 23.8% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sepuluh perpuluhan enam megabit sesaat, iaitu peningkatan sebanyak dua puluh tiga perpuluhan lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 91.2 Mbps, iaitu peningkatan sebanyak 80% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah sembilan puluh satu perpuluhan dua megabit sesaat, iaitu peningkatan sebanyak lapan puluh peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 40. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal empat puluh ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 14.14 Mbps, iaitu peningkatan sebanyak 44% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah empat belas perpuluhan satu empat megabit sesaat, iaitu peningkatan sebanyak empat puluh empat peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 62.75 Mbps, iaitu peningkatan sebanyak 86% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah enam puluh dua perpuluhan tujuh lima megabit sesaat, iaitu peningkatan sebanyak lapan puluh enam peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan PO 55219 anda sedang diproses dan akan dihantar dalam masa 812 hari bekerja."}, {"role": "assistant", "content": "Pesanan PO lima lima dua satu sembilan anda sedang diproses dan akan dihantar dalam masa lapan ratus dua belas hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 0.1 Mbps, iaitu peningkatan sebanyak 46% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah kosong perpuluhan satu megabit sesaat, iaitu peningkatan sebanyak empat puluh enam peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 13229. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah satu tiga dua dua sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 7689. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah tujuh enam lapan sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod OTP anda ialah 1621. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod OTP anda ialah satu enam dua satu. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 86.26 Mbps, iaitu peningkatan sebanyak 86.8% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lapan puluh enam perpuluhan dua enam megabit sesaat, iaitu peningkatan sebanyak lapan puluh enam perpuluhan lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 867989. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lapan enam tujuh sembilan lapan sembilan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 18.17 Mbps, iaitu peningkatan sebanyak 78% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah lapan belas perpuluhan satu tujuh megabit sesaat, iaitu peningkatan sebanyak tujuh puluh lapan peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan X-12340567 anda sedang diproses dan akan dihantar dalam masa 56 hari bekerja."}, {"role": "assistant", "content": "Pesanan X satu dua tiga empat kosong lima enam tujuh anda sedang diproses dan akan dihantar dalam masa lima puluh enam hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM745. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal tujuh ratus empat puluh lima ringgit. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pesanan INV0472 anda sedang diproses dan akan dihantar dalam masa 1 hari bekerja."}, {"role": "assistant", "content": "Pesanan INV kosong empat tujuh dua anda sedang diproses dan akan dihantar dalam masa satu hari bekerja."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kelajuan internet anda kini mencecah 41.4 Mbps, iaitu peningkatan sebanyak 62% daripada bulan lepas."}, {"role": "assistant", "content": "Kelajuan internet anda kini mencecah empat puluh satu perpuluhan empat megabit sesaat, iaitu peningkatan sebanyak enam puluh dua peratus daripada bulan lepas."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 3388. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah tiga tiga lapan lapan. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Baki akaun anda kini tinggal RM 25.12. Sila tambah nilai segera."}, {"role": "assistant", "content": "Baki akaun anda kini tinggal dua puluh lima ringgit dua belas sen. Sila tambah nilai segera."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 1818. Jangan kongsi kod ini dengan sesiapa."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah seribu lapan ratus lapan belas. Jangan kongsi kod ini dengan sesiapa."}], "lang": "ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1 °C."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu derajat Celsius."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 306gr."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga ratus enam gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 55957, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah lima lima sembilan lima tujuh, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 26/06/2010 pukul 18."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh enam Juni dua ribu sepuluh pukul delapan belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah ABC1234, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah ABC satu dua tiga empat, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah MH370, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah MH tiga tujuh nol, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 4,9ml."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat koma sembilan mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1 gr."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 332 cm."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga ratus tiga puluh dua sentimeter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 17-11-2031 pukul 19:00."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada tujuh belas November dua ribu tiga puluh satu pukul sembilan belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 25/06/2009 pukul 10."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh lima Juni dua ribu sembilan pukul sepuluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 302 kg."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga ratus dua kilogram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1,9 g."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu koma sembilan gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 02248, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol dua dua empat delapan, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 426,1 GB."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat ratus dua puluh enam koma satu gigabyte."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah REF-2024-0113, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah REF dua nol dua empat nol satu satu tiga, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 01213, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol satu dua satu tiga, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah B2-14, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah B dua satu empat, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 48362, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah empat delapan tiga enam dua, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 8 GB."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal delapan gigabyte."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 25/03/2017 pukul 4.05."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh lima Maret dua ribu tujuh belas pukul empat nol lima."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 497,2 gr."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat ratus sembilan puluh tujuh koma dua gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1 m."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu meter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah QZ8501, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah QZ delapan lima nol satu, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 4 Agustus pukul 7.30."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada empat Agustus pukul tujuh tiga puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 02-07-2017 pukul 4.00."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua Juli dua ribu tujuh belas pukul empat."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 611095, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah enam satu satu nol sembilan lima, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 6,2 cm."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal enam koma dua sentimeter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 14/12/2018 pukul 12.20."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada empat belas Desember dua ribu delapan belas pukul dua belas dua puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1,8 l."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu koma delapan liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 9432, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah sembilan empat tiga dua, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1ml."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 206cm."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal dua ratus enam sentimeter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah PO 55219, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah PO lima lima dua satu sembilan, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 6,5g."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal enam koma lima gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 18/05/2019 pukul 16."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada delapan belas Mei dua ribu sembilan belas pukul enam belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 20 Januari 2010 pukul 16.10."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh Januari dua ribu sepuluh pukul enam belas sepuluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 9l."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal sembilan liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 3,7 l."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga koma tujuh liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah A12, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah A satu dua, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 24 Juni 2008 pukul 8:20."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh empat Juni dua ribu delapan pukul delapan dua puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 11/08/1999 pukul 9.10."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada sebelas Agustus seribu sembilan ratus sembilan puluh sembilan pukul sembilan sepuluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 9 Januari 2007 pukul 8:15."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada sembilan Januari dua ribu tujuh pukul delapan lima belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 15 Januari pukul 22:00."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada lima belas Januari pukul dua puluh dua."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 3L."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah INV0472, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah INV nol empat tujuh dua, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 08264, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol delapan dua enam empat, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 909870, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah sembilan nol sembilan delapan tujuh nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 6 ml."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal enam mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 7 Desember 2021 pukul 7.05."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada tujuh Desember dua ribu dua puluh satu pukul tujuh nol lima."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1cm."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu sentimeter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 3,1m."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tiga koma satu meter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 6195, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah enam satu sembilan lima, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 475L."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat ratus tujuh puluh lima liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 19-03-2031 pukul 8.30."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada sembilan belas Maret dua ribu tiga puluh satu pukul delapan tiga puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 1910, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah satu sembilan satu nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 2 ml."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal dua mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 12-07-1998 pukul 3:00."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua belas Juli seribu sembilan ratus sembilan puluh delapan pukul tiga."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 27 April 2026 pukul 3.50."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh tujuh April dua ribu dua puluh enam pukul tiga lima puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 7kg."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tujuh kilogram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah AK6120, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah AK enam satu dua nol, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 06-07-2017 pukul 3:00."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada enam Juli dua ribu tujuh belas pukul tiga."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1 °C."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu derajat Celsius."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 20 Mei pukul 0."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh Mei pukul nol."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 22-09-2031 pukul 10:50."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh dua September dua ribu tiga puluh satu pukul sepuluh lima puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 62,5 km."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal enam puluh dua koma lima kilometer."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1 L."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu liter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 365 derajat."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal tiga ratus enam puluh lima derajat Celsius."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nomor referensi laporan Anda adalah X-12340567, mohon simpan untuk pengecekan."}, {"role": "assistant", "content": "Nomor referensi laporan Anda adalah X satu dua tiga empat nol lima enam tujuh, mohon simpan untuk pengecekan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 85780, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah delapan lima tujuh delapan nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 0651, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol enam lima satu, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 076815, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol tujuh enam delapan satu lima, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 00833, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol nol delapan tiga tiga, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 4/10/1999 pukul 3.15."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada empat Oktober seribu sembilan ratus sembilan puluh sembilan pukul tiga lima belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 0771, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol tujuh tujuh satu, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 18 Agustus 2020 pukul 9:40."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada delapan belas Agustus dua ribu dua puluh pukul sembilan empat puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 0639, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah nol enam tiga sembilan, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 25 Juni pukul 7.30."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh lima Juni pukul tujuh tiga puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 6371, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah enam tiga tujuh satu, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 78 km."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal tujuh puluh delapan kilometer."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1,7kg."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu koma tujuh kilogram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 4632, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah empat enam tiga dua, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 292ml."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal dua ratus sembilan puluh dua mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 315777, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah tiga satu lima tujuh tujuh tujuh, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 9372, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah sembilan tiga tujuh dua, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 12 Agustus 1998 pukul 0.20."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua belas Agustus seribu sembilan ratus sembilan puluh delapan pukul nol dua puluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 406,8 ml."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat ratus enam koma delapan mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 6/10/2020 pukul 0."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada enam Oktober dua ribu dua puluh pukul nol."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 2 gr."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal dua gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 81570, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah delapan satu lima tujuh nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1,8 °C."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu koma delapan derajat Celsius."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 8077, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah delapan nol tujuh tujuh, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 7170, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah tujuh satu tujuh nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1g."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 37758, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah tiga tujuh tujuh lima delapan, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 14/07/2020 pukul 13:10."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada empat belas Juli dua ribu dua puluh pukul tiga belas sepuluh."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 12 Juni 2017 pukul 8:05."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua belas Juni dua ribu tujuh belas pukul delapan nol lima."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 8 gr."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal delapan gram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 3870, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah tiga delapan tujuh nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 98774, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah sembilan delapan tujuh tujuh empat, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 5 Mei pukul 9."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada lima Mei pukul sembilan."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 2cm."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal dua sentimeter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 244738, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah dua empat empat tujuh tiga delapan, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 120993, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah satu dua nol sembilan sembilan tiga, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 6434, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah enam empat tiga empat, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 2,7 GB."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal dua koma tujuh gigabyte."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 1kg."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal satu kilogram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 175 km."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal seratus tujuh puluh lima kilometer."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 1,1 ml."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal satu koma satu mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal 485kg."}, {"role": "assistant", "content": "Mohon unggah berkas dalam format PDF dengan ukuran maksimal empat ratus delapan puluh lima kilogram."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 8497, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah delapan empat sembilan tujuh, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sistem kami akan melakukan pemeliharaan rutin pada 23/01/2017 pukul 19."}, {"role": "assistant", "content": "Sistem kami akan melakukan pemeliharaan rutin pada dua puluh tiga Januari dua ribu tujuh belas pukul sembilan belas."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal 4 ml."}, {"role": "assistant", "content": "Mohon unggah dokumen pendukung dengan ukuran maksimal empat mililiter."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 27802, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah dua tujuh delapan nol dua, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 504890, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah lima nol empat delapan sembilan nol, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kode OTP Anda adalah 2387, mohon jangan berikan kode ini kepada siapa pun."}, {"role": "assistant", "content": "Kode OTP Anda adalah dua tiga delapan tujuh, mohon jangan berikan kode ini kepada siapa pun."}], "lang": "id", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 14.7%，建议您前往 https://sltnet.lk/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之十四点七，建议您前往 h t t p s colon slash slash sltnet dot lk slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 83.4%，建议您前往 https://www.gmail.co.id/support 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之八十三点四，建议您前往 h t t p s colon slash slash www dot gmail dot co dot id slash support 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM65.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为六十五令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至014-988 7820以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一四九八八七八二零以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 21%，建议您前往 www.gmail.co.id 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之二十一，建议您前往 w w w dot gmail dot co dot id 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM392.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三百九十二令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 maria71@onet.pl。"}, {"role": "assistant", "content": "请将相关证明文件发送至 maria 七一 at onet dot pl。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至010-731 2531以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一零七三一二五三一以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 03-0394 6308。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零三零三九四六三零八。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 011-48019628。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一一四八零一九六二八。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 03-5317 5563。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零三五三一七五五六三。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 32.85，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三十二令吉八十五仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM202.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为二百零二令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 65%，建议您前往 https://correo.es 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之六十五，建议您前往 h t t p s colon slash slash correo dot es 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 ravi52@web.de。"}, {"role": "assistant", "content": "请将相关证明文件发送至 ravi 五二 at web dot de。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 luca@ziggo.nl。"}, {"role": "assistant", "content": "请将相关证明文件发送至 luca at ziggo dot nl。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM20,822，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为二万零八百二十二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 012-986 0574。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一二九八六零五七四。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至014-700 1690以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一四七零零一六九零以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 52%，建议您前往 https://www.yahoo.com/support 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之五十二，建议您前往 h t t p s colon slash slash www dot yahoo dot com slash support 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 maria1@web.de。"}, {"role": "assistant", "content": "请将相关证明文件发送至 maria 一 at web dot de。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 14-568 6870。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一四五六八六八七零。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM75，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为七十五令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 40%，建议您前往 www.ziggo.nl 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之四十，建议您前往 w w w dot ziggo dot nl 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 59%，建议您前往 www.web.de 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之五十九，建议您前往 w w w dot web dot de 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM48.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为四十八令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 77.2%，建议您前往 https://yahoo.com/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之七十七点二，建议您前往 h t t p s colon slash slash yahoo dot com slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM35,661.60，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三万五千六百六十一令吉六十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 78,562，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为七万八千五百六十二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至03-7155 4412以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零三七一五五四四一二以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 jean.dupont@onet.pl。"}, {"role": "assistant", "content": "请将相关证明文件发送至 jean dot dupont at onet dot pl。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 6%，建议您前往 https://onet.pl/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之六，建议您前往 h t t p s colon slash slash onet dot pl slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 31%，建议您前往 https://web.de 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之三十一，建议您前往 h t t p s colon slash slash web dot de 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 36%，建议您前往 https://company.co/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之三十六，建议您前往 h t t p s colon slash slash company dot co slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 10,257.61，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为一万零二百五十七令吉六十一仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 dilani88@example.com。"}, {"role": "assistant", "content": "请将相关证明文件发送至 dilani 八八 at example dot com。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至013-738 7046以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一三七三八七零四六以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 015-858 6856。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一五八五八六八五六。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM957.27，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为九百五十七令吉二十七仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至018-525 1627以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一八五二五一六二七以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 jan.devries@mail.fr。"}, {"role": "assistant", "content": "请将相关证明文件发送至 jan dot devries at mail dot fr。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 support@sapo.pt。"}, {"role": "assistant", "content": "请将相关证明文件发送至 support at sapo dot pt。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM10,534.20，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为一万零五百三十四令吉二十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 17-818 8728。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一七八一八八七二八。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至03-9295 1754以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零三九二九五一七五四以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至012-00900330以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一二零零九零零三三零以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM933.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为九百三十三令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 ahmad.b@mail.fr。"}, {"role": "assistant", "content": "请将相关证明文件发送至 ahmad dot b at mail dot fr。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 ali@bank.my。"}, {"role": "assistant", "content": "请将相关证明文件发送至 ali at bank dot my。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 luca@yahoo.com。"}, {"role": "assistant", "content": "请将相关证明文件发送至 luca at yahoo dot com。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 011-112 1615。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一一一一二一六一五。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 49%，建议您前往 www.ziggo.nl/login 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之四十九，建议您前往 w w w dot ziggo dot nl slash login 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 489，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为四百八十九令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 17-994 6676。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一七九九四六六七六。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 018-03753053。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一八零三七五三零五三。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 019-524 7243。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一九五二四七二四三。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM15，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为十五令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 91%，建议您前往 https://correo.es 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之九十一，建议您前往 h t t p s colon slash slash correo dot es 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM66，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为六十六令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 36%，建议您前往 www.sapo.pt/login 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之三十六，建议您前往 w w w dot sapo dot pt slash login 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM37,317.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三万七千三百一十七令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至+60 15-287 8481以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至加六零一五二八七八四八一以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 11-244 1929。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一一二四四一九二九。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 29%，建议您前往 https://www.correo.es/support 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之二十九，建议您前往 h t t p s colon slash slash www dot correo dot es slash support 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 4%，建议您前往 www.ziggo.nl/login 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之四，建议您前往 w w w dot ziggo dot nl slash login 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 10.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为十令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 015-03847807。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一五零三八四七八零七。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 chen88@example.com。"}, {"role": "assistant", "content": "请将相关证明文件发送至 chen 八八 at example dot com。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM5,290，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为五千二百九十令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM34,482，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三万四千四百八十二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 49.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为四十九令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 016-75575459。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一六七五五七五四五九。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 84%，建议您前往 https://yahoo.com.ph 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之八十四，建议您前往 h t t p s colon slash slash yahoo dot com dot ph 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至015-261 1994以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一五二六一一九九四以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至017-500 5450以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一七五零零五四五零以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 info23@yahoo.com.ph。"}, {"role": "assistant", "content": "请将相关证明文件发送至 info 二三 at yahoo dot com dot ph。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 21%，建议您前往 https://gmail.com/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之二十一，建议您前往 h t t p s colon slash slash gmail dot com slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 mei.ling@sltnet.lk。"}, {"role": "assistant", "content": "请将相关证明文件发送至 mei dot ling at sltnet dot lk。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM82，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为八十二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 60%，建议您前往 https://example.com/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之六十，建议您前往 h t t p s colon slash slash example dot com slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 80%，建议您前往 https://www.ziggo.nl/support 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之八十，建议您前往 h t t p s colon slash slash www dot ziggo dot nl slash support 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 33%，建议您前往 https://yahoo.com/help 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之三十三，建议您前往 h t t p s colon slash slash yahoo dot com slash help 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至03-4086 0182以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零三四零八六零一八二以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 kasia96@posta.it。"}, {"role": "assistant", "content": "请将相关证明文件发送至 kasia 九六 at posta dot it。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 011-18991292。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一一一八九九一二九二。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 03-1625 0985。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零三一六二五零九八五。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 ali67@sltnet.lk。"}, {"role": "assistant", "content": "请将相关证明文件发送至 ali 六七 at sltnet dot lk。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 529.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为五百二十九令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 15-067 1548。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一五零六七一五四八。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 18-824 2593。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一八八二四二五九三。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 62%，建议您前往 https://ziggo.nl/faq 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之六十二，建议您前往 h t t p s colon slash slash ziggo dot nl slash faq 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 366.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三百六十六令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 help@ziggo.nl。"}, {"role": "assistant", "content": "请将相关证明文件发送至 help at ziggo dot nl。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM11，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为十一令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 support97@web.de。"}, {"role": "assistant", "content": "请将相关证明文件发送至 support 九七 at web dot de。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM92,038.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为九万二千零三十八令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 9.4%，建议您前往 www.onet.pl/login 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之九点四，建议您前往 w w w dot onet dot pl slash login 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 93.72，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为九十三令吉七十二仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 95%，建议您前往 https://outlook.sa 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之九十五，建议您前往 h t t p s colon slash slash outlook dot sa 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 015-52513100。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一五五二五一三一零零。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 012-71259715。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一二七一二五九七一五。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 7,294，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为七千二百九十四令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 38%，建议您前往 https://bank.my/faq 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之三十八，建议您前往 h t t p s colon slash slash bank dot my slash faq 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 645，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为六百四十五令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 24,293.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为二万四千二百九十三令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至011-10561256以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一一一零五六一二五六以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 help@onet.pl。"}, {"role": "assistant", "content": "请将相关证明文件发送至 help at onet dot pl。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 pedro@outlook.sa。"}, {"role": "assistant", "content": "请将相关证明文件发送至 pedro at outlook dot sa。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 402，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为四百零二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 juan@yahoo.com.ph。"}, {"role": "assistant", "content": "请将相关证明文件发送至 juan at yahoo dot com dot ph。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 017-890 9936。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一七八九零九九三六。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 03-2459 3097。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零三二四五九三零九七。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 03-1917 5350。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零三一九一七五三五零。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 49,657，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为四万九千六百五十七令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 69%，建议您前往 https://www.ziggo.nl/support 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之六十九，建议您前往 h t t p s colon slash slash www dot ziggo dot nl slash support 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请将相关证明文件发送至 ahmad.b@company.co。"}, {"role": "assistant", "content": "请将相关证明文件发送至 ahmad dot b at company dot co。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 83,683，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为八万三千六百八十三令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 13-072 4185。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一三零七二四一八五。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM 11.90，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为十一令吉九十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 018-09263581。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一八零九二六三五八一。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至019-199 3205以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至零一九一九九三二零五以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM32.50，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为三十二令吉五十仙，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的流量包已使用 59.4%，建议您前往 www.web.de/login 办理升级。"}, {"role": "assistant", "content": "您的流量包已使用 百分之五十九点四，建议您前往 w w w dot web dot de slash login 办理升级。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 +60 16-601 9967。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 加六零一六六零一九九六七。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您有任何疑问，请拨打招生办电话 017-239 1690。"}, {"role": "assistant", "content": "如果您有任何疑问，请拨打招生办电话 零一七二三九一六九零。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额为RM72，请及时充值以继续使用服务。"}, {"role": "assistant", "content": "您的账户余额为七十二令吉，请及时充值以继续使用服务。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请发送您的身份证号至+60 11-439 5697以办理实名认证。"}, {"role": "assistant", "content": "请发送您的身份证号至加六零一一四三九五六九七以办理实名认证。"}], "lang": "zh", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 24 - 29 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்து நான்கு முதல் இருபத்தொன்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 6 அக்டோபர் 2023 அன்று 1:00 p.m. மணிக்கு, அறை 601025."}, {"role": "assistant", "content": "கூட்டம் ஆறு அக்டோபர் இரண்டாயிரத்து இருபத்து மூன்று அன்று ஒன்று பி எம் மணிக்கு, அறை ஆறு பூஜ்ஜியம் ஒன்று பூஜ்ஜியம் இரண்டு ஐந்து."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 13 - 22 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதிமூன்று முதல் இருபத்திரண்டு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 15–22 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினைந்து முதல் இருபத்திரண்டு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் ஜூன் 21 அன்று 23:20 மணிக்கு, அறை 63164."}, {"role": "assistant", "content": "கூட்டம் ஜூன் இருபத்தொன்று அன்று இருபத்து மூன்று இருபது மணிக்கு, அறை ஆறு மூன்று ஒன்று ஆறு நான்கு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 5 ஜூலை 1999 அன்று 9:15 p.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஐந்து ஜூலை ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஒன்பது அன்று ஒன்பது பதினைந்து பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 7–14 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை ஏழு முதல் பதினான்கு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 10–17 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பத்து முதல் பதினேழு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 18 நவம்பர் 2000 அன்று 12am மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினெட்டு நவம்பர் இரண்டாயிரம் அன்று பன்னிரண்டு ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 16 ஜூலை 2021 அன்று 2a.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினாறு ஜூலை இரண்டாயிரத்து இருபத்தொன்று அன்று இரண்டு ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 253 GB ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு இருநூற்று ஐம்பத்து மூன்று ஜிபி ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 18-23 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினெட்டு முதல் இருபத்து மூன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 24-26 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்து நான்கு முதல் இருபத்தாறு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 15/8/2027 அன்று 6pm மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினைந்து ஆகஸ்ட் இரண்டாயிரத்து இருபத்தேழு அன்று ஆறு பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 20–28 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபது முதல் இருபத்தெட்டு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 11/10/1997 அன்று 4:20 a.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினொன்று அக்டோபர் ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஏழு அன்று நான்கு இருபது ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் டிசம்பர் 13 அன்று 8.45 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் டிசம்பர் பதிமூன்று அன்று எட்டு நாற்பத்தைந்து மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 111km ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நூற்று பதினொன்று கிலோமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 26/2/2017 அன்று 14:40 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இருபத்தாறு பிப்ரவரி இரண்டாயிரத்து பதினேழு அன்று பதினான்கு நாற்பது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 102 ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நூற்று இரண்டு மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2018-10-12 அன்று 5.00 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பன்னிரண்டு அக்டோபர் இரண்டாயிரத்து பதினெட்டு அன்று ஐந்து மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2003-02-12 அன்று 14:00 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பன்னிரண்டு பிப்ரவரி இரண்டாயிரத்து மூன்று அன்று பதினான்கு மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 45.1kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நாற்பத்தைந்து புள்ளி ஒன்று கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 13 - 19 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதிமூன்று முதல் பத்தொன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 1999-01-07 அன்று 11.00 மணிக்கு, அறை 491405."}, {"role": "assistant", "content": "கூட்டம் ஏழு ஜனவரி ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஒன்பது அன்று பதினொன்று மணிக்கு, அறை நான்கு ஒன்பது ஒன்று நான்கு பூஜ்ஜியம் ஐந்து."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 40g ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நாற்பது கிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 15 அக்டோபர் 2010 அன்று 7:40 AM மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினைந்து அக்டோபர் இரண்டாயிரத்து பத்து அன்று ஏழு நாற்பது ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 27-30 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்தேழு முதல் முப்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 15–18 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினைந்து முதல் பதினெட்டு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 4.5 km ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நான்கு புள்ளி ஐந்து கிலோமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 51km ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு ஐம்பத்தொன்று கிலோமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 23-29 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபத்து மூன்று முதல் இருபத்தொன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 218 l ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு இருநூற்று பதினெட்டு லிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 80 GB ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எண்பது ஜிபி ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 25-33 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபத்தைந்து முதல் முப்பத்து மூன்று மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 11/1/2022 அன்று 4:45 p.m. மணிக்கு, அறை 797628."}, {"role": "assistant", "content": "கூட்டம் பதினொன்று ஜனவரி இரண்டாயிரத்து இருபத்திரண்டு அன்று நான்கு நாற்பத்தைந்து பி எம் மணிக்கு, அறை ஏழு ஒன்பது ஏழு ஆறு இரண்டு எட்டு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை 19 அன்று 6:30 am மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை பத்தொன்பது அன்று ஆறு முப்பது ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 17–22 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினேழு முதல் இருபத்திரண்டு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 4/9/2032 அன்று 9:00 p.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் நான்கு செப்டம்பர் இரண்டாயிரத்து முப்பத்திரண்டு அன்று ஒன்பது பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 87.1 °C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எண்பத்தேழு புள்ளி ஒன்று டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 14/8/2015 அன்று 9.00 மணிக்கு, அறை 34421."}, {"role": "assistant", "content": "கூட்டம் பதினான்கு ஆகஸ்ட் இரண்டாயிரத்து பதினைந்து அன்று ஒன்பது மணிக்கு, அறை மூன்று நான்கு நான்கு இரண்டு ஒன்று."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 43.7 g ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நாற்பத்து மூன்று புள்ளி ஏழு கிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 4-5 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு நான்கு முதல் ஐந்து வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 10 ஜூலை 2023 அன்று 2.45 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பத்து ஜூலை இரண்டாயிரத்து இருபத்து மூன்று அன்று இரண்டு நாற்பத்தைந்து மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 66.1ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு அறுபத்தாறு புள்ளி ஒன்று மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜனவரி 22 அன்று 9.20 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜனவரி இருபத்திரண்டு அன்று ஒன்பது இருபது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 389 GB ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு முந்நூற்று எண்பத்தொன்பது ஜிபி ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 10.8 l ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு பத்து புள்ளி எட்டு லிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 17-19 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினேழு முதல் பத்தொன்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 16 - 23 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினாறு முதல் இருபத்து மூன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2000-01-22 அன்று 11.00 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இருபத்திரண்டு ஜனவரி இரண்டாயிரம் அன்று பதினொன்று மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 16 - 19 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினாறு முதல் பத்தொன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2000-03-16 அன்று 11:00 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பதினாறு மார்ச் இரண்டாயிரம் அன்று பதினொன்று மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 12/5/2002 அன்று 2.50 மணிக்கு, அறை 903238."}, {"role": "assistant", "content": "கூட்டம் பன்னிரண்டு மே இரண்டாயிரத்து இரண்டு அன்று இரண்டு ஐம்பது மணிக்கு, அறை ஒன்பது பூஜ்ஜியம் மூன்று இரண்டு மூன்று எட்டு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 8.7 kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எட்டு புள்ளி ஏழு கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 17-21 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினேழு முதல் இருபத்தொன்று மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை 5 அன்று 16:20 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை ஐந்து அன்று பதினாறு இருபது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 2010-06-23 அன்று 4p.m. மணிக்கு, அறை 591474."}, {"role": "assistant", "content": "கூட்டம் இருபத்து மூன்று ஜூன் இரண்டாயிரத்து பத்து அன்று நான்கு பி எம் மணிக்கு, அறை ஐந்து ஒன்பது ஒன்று நான்கு ஏழு நான்கு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 133ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நூற்று முப்பத்து மூன்று மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 19 - 25 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பத்தொன்பது முதல் இருபத்தைந்து வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 5–9 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை ஐந்து முதல் ஒன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 25/11/2010 அன்று 1.10 மணிக்கு, அறை 58688."}, {"role": "assistant", "content": "கூட்டம் இருபத்தைந்து நவம்பர் இரண்டாயிரத்து பத்து அன்று ஒன்று பத்து மணிக்கு, அறை ஐந்து எட்டு ஆறு எட்டு எட்டு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 68°C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு அறுபத்தெட்டு டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 26 - 30 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்தாறு முதல் முப்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 3-10 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை மூன்று முதல் பத்து மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 19–25 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பத்தொன்பது முதல் இருபத்தைந்து வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 16–21 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினாறு முதல் இருபத்தொன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2024-02-01 அன்று 1a.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஒன்று பிப்ரவரி இரண்டாயிரத்து இருபத்து நான்கு அன்று ஒன்று ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 3 ஜூன் 2032 அன்று 4:30 am மணிக்கு, அறை 959214."}, {"role": "assistant", "content": "கூட்டம் மூன்று ஜூன் இரண்டாயிரத்து முப்பத்திரண்டு அன்று நான்கு முப்பது ஏ எம் மணிக்கு, அறை ஒன்பது ஐந்து ஒன்பது இரண்டு ஒன்று நான்கு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 13-23 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதிமூன்று முதல் இருபத்து மூன்று மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 16-19 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினாறு முதல் பத்தொன்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 92.3kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு தொண்ணூற்று இரண்டு புள்ளி மூன்று கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2025-06-23 அன்று 6:15 AM மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இருபத்து மூன்று ஜூன் இரண்டாயிரத்து இருபத்தைந்து அன்று ஆறு பதினைந்து ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 345 ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு முந்நூற்று நாற்பத்தைந்து மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 28.3 kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு இருபத்தெட்டு புள்ளி மூன்று கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 20-23 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபது முதல் இருபத்து மூன்று மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 1–10 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு ஒன்று முதல் பத்து வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 1/3/2000 அன்று 6.30 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஒன்று மார்ச் இரண்டாயிரம் அன்று ஆறு முப்பது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 3/8/2017 அன்று 9pm மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் மூன்று ஆகஸ்ட் இரண்டாயிரத்து பதினேழு அன்று ஒன்பது பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் செப்டம்பர் 3 அன்று 5.40 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் செப்டம்பர் மூன்று அன்று ஐந்து நாற்பது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 20/1/2025 அன்று 8:20 a.m. மணிக்கு, அறை 48040."}, {"role": "assistant", "content": "கூட்டம் இருபது ஜனவரி இரண்டாயிரத்து இருபத்தைந்து அன்று எட்டு இருபது ஏ எம் மணிக்கு, அறை நான்கு எட்டு பூஜ்ஜியம் நான்கு பூஜ்ஜியம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 24-33 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபத்து நான்கு முதல் முப்பத்து மூன்று மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 280ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு இருநூற்று எண்பது மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 9 - 15 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு ஒன்பது முதல் பதினைந்து வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 431 ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நானூற்று முப்பத்தொன்று மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 7-16 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை ஏழு முதல் பதினாறு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 4.4°C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நான்கு புள்ளி நான்கு டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 93.6ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு தொண்ணூற்று மூன்று புள்ளி ஆறு மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் மே 16 அன்று 3.50 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் மே பதினாறு அன்று மூன்று ஐம்பது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 6-8 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு ஆறு முதல் எட்டு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை 19 அன்று 22:00 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை பத்தொன்பது அன்று இருபத்திரண்டு மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 14 செப்டம்பர் 2032 அன்று 1pm மணிக்கு, அறை 1433."}, {"role": "assistant", "content": "கூட்டம் பதினான்கு செப்டம்பர் இரண்டாயிரத்து முப்பத்திரண்டு அன்று ஒன்று பி எம் மணிக்கு, அறை ஒன்று நான்கு மூன்று மூன்று."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 7-14 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை ஏழு முதல் பதினான்கு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 88.9g ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எண்பத்தெட்டு புள்ளி ஒன்பது கிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 49 °C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நாற்பத்தொன்பது டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 8-15 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை எட்டு முதல் பதினைந்து மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2020-05-02 அன்று 10:00 p.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இரண்டு மே இரண்டாயிரத்து இருபது அன்று பத்து பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் மார்ச் 3 அன்று 9p.m. மணிக்கு, அறை 68877."}, {"role": "assistant", "content": "கூட்டம் மார்ச் மூன்று அன்று ஒன்பது பி எம் மணிக்கு, அறை ஆறு எட்டு எட்டு ஏழு ஏழு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 32.3l ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு முப்பத்திரண்டு புள்ளி மூன்று லிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 61 °C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு அறுபத்தொன்று டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 16 - 24 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு பதினாறு முதல் இருபத்து நான்கு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 9 - 17 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை ஒன்பது முதல் பதினேழு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் ஜனவரி 1 அன்று 17:40 மணிக்கு, அறை 88306."}, {"role": "assistant", "content": "கூட்டம் ஜனவரி ஒன்று அன்று பதினேழு நாற்பது மணிக்கு, அறை எட்டு எட்டு மூன்று பூஜ்ஜியம் ஆறு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 18.4cm ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு பதினெட்டு புள்ளி நான்கு சென்டிமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 108GB ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நூற்று எட்டு ஜிபி ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 2.8kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு இரண்டு புள்ளி எட்டு கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 9-19 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு ஒன்பது முதல் பத்தொன்பது வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 2002-04-07 அன்று 2:00 PM மணிக்கு, அறை 3444."}, {"role": "assistant", "content": "கூட்டம் ஏழு ஏப்ரல் இரண்டாயிரத்து இரண்டு அன்று இரண்டு பி எம் மணிக்கு, அறை மூன்று நான்கு நான்கு நான்கு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 144 °C ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு நூற்று நாற்பத்து நான்கு டிகிரி செல்சியஸ் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 70.4 km ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எழுபது புள்ளி நான்கு கிலோமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 9-11 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு ஒன்பது முதல் பதினொன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் செப்டம்பர் 17 அன்று 11PM மணிக்கு, அறை 97499."}, {"role": "assistant", "content": "கூட்டம் செப்டம்பர் பதினேழு அன்று பதினொன்று பி எம் மணிக்கு, அறை ஒன்பது ஏழு நான்கு ஒன்பது ஒன்பது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 20 டிசம்பர் 2016 அன்று 6:15 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இருபது டிசம்பர் இரண்டாயிரத்து பதினாறு அன்று ஆறு பதினைந்து மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 28 - 36 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபத்தெட்டு முதல் முப்பத்தாறு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 17-19 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினேழு முதல் பத்தொன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் நவம்பர் 7 அன்று 7pm மணிக்கு, அறை 16012."}, {"role": "assistant", "content": "கூட்டம் நவம்பர் ஏழு அன்று ஏழு பி எம் மணிக்கு, அறை ஒன்று ஆறு பூஜ்ஜியம் ஒன்று இரண்டு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 13 ஆகஸ்ட் 2023 அன்று 6:15 AM மணிக்கு, அறை 32551."}, {"role": "assistant", "content": "கூட்டம் பதிமூன்று ஆகஸ்ட் இரண்டாயிரத்து இருபத்து மூன்று அன்று ஆறு பதினைந்து ஏ எம் மணிக்கு, அறை மூன்று இரண்டு ஐந்து ஐந்து ஒன்று."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 15-16 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினைந்து முதல் பதினாறு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை 1 அன்று 3p.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஜூலை ஒன்று அன்று மூன்று பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2020-07-02 அன்று 7:40 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இரண்டு ஜூலை இரண்டாயிரத்து இருபது அன்று ஏழு நாற்பது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 62ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு அறுபத்திரண்டு மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 3–8 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை மூன்று முதல் எட்டு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 2025-03-20 அன்று 8.40 மணிக்கு, அறை 431871."}, {"role": "assistant", "content": "கூட்டம் இருபது மார்ச் இரண்டாயிரத்து இருபத்தைந்து அன்று எட்டு நாற்பது மணிக்கு, அறை நான்கு மூன்று ஒன்று எட்டு ஏழு ஒன்று."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 8–16 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு எட்டு முதல் பதினாறு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 18–20 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பதினெட்டு முதல் இருபது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 26–27 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்தாறு முதல் இருபத்தேழு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 26-34 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்தாறு முதல் முப்பத்து நான்கு வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 339 kg ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு முந்நூற்று முப்பத்தொன்பது கிலோகிராம் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 26 ஜூன் 2029 அன்று 9a.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் இருபத்தாறு ஜூன் இரண்டாயிரத்து இருபத்தொன்பது அன்று ஒன்பது ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 13 cm ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு பதிமூன்று சென்டிமீட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 2032-03-03 அன்று 12:20 PM மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் மூன்று மார்ச் இரண்டாயிரத்து முப்பத்திரண்டு அன்று பன்னிரண்டு இருபது பி எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஏப்ரல் 14 அன்று 7a.m. மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் ஏப்ரல் பதினான்கு அன்று ஏழு ஏ எம் மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 4 ஏப்ரல் 2013 அன்று 7:30 மணிக்கு, அறை 74493."}, {"role": "assistant", "content": "கூட்டம் நான்கு ஏப்ரல் இரண்டாயிரத்து பதிமூன்று அன்று ஏழு முப்பது மணிக்கு, அறை ஏழு நான்கு நான்கு ஒன்பது மூன்று."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 12 ஜூலை 2016 அன்று 0:15 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பன்னிரண்டு ஜூலை இரண்டாயிரத்து பதினாறு அன்று பூஜ்ஜியம் பதினைந்து மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 4-11 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு நான்கு முதல் பதினொன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் 10 ஜூலை 2010 அன்று 6.20 மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}, {"role": "assistant", "content": "உங்கள் அப்பாயிண்ட்மெண்ட் பத்து ஜூலை இரண்டாயிரத்து பத்து அன்று ஆறு இருபது மணிக்கு உறுதி செய்யப்பட்டுள்ளது."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 12 - 18 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பன்னிரண்டு முதல் பதினெட்டு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 1.5ml ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு ஒரு புள்ளி ஐந்து மில்லிலிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த பரிவர்த்தனைக்கு 23–33 வேலை நாட்கள் ஆகும்."}, {"role": "assistant", "content": "இந்த பரிவர்த்தனைக்கு இருபத்து மூன்று முதல் முப்பத்து மூன்று வேலை நாட்கள் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 9 மார்ச் 2012 அன்று 5p.m. மணிக்கு, அறை 8580."}, {"role": "assistant", "content": "கூட்டம் ஒன்பது மார்ச் இரண்டாயிரத்து பன்னிரண்டு அன்று ஐந்து பி எம் மணிக்கு, அறை எட்டு ஐந்து எட்டு பூஜ்ஜியம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 12-22 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை பன்னிரண்டு முதல் இருபத்திரண்டு மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு 76.9 l ஆகும்."}, {"role": "assistant", "content": "இந்த மாதத்திற்கான மொத்தப் பயன்பாடு எழுபத்தாறு புள்ளி ஒன்பது லிட்டர் ஆகும்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நாங்கள் காலை 24 - 29 மணி வரை இயங்குவோம்."}, {"role": "assistant", "content": "நாங்கள் காலை இருபத்து நான்கு முதல் இருபத்தொன்பது மணி வரை இயங்குவோம்."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கூட்டம் 2008-12-23 அன்று 3am மணிக்கு, அறை 4252."}, {"role": "assistant", "content": "கூட்டம் இருபத்து மூன்று டிசம்பர் இரண்டாயிரத்து எட்டு அன்று மூன்று ஏ எம் மணிக்கு, அறை நான்கு இரண்டு ஐந்து இரண்டு."}], "lang": "ta", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 319617 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட மூன்று லட்சத்து பத்தொன்பதாயிரத்து அறுநூற்று பதினேழு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 79 646 9390 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஒன்பது ஆறு நான்கு ஆறு ஒன்பது மூன்று ஒன்பது பூஜ்ஜியம் என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது LKR 437.30 ஆகும், இதனை நவம்பர் 3 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது நானூற்று முப்பத்தேழு ரூபாய் முப்பது சதம் ஆகும், இதனை நவம்பர் மூன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை help@sltnet.lk க்கு அனுப்புங்கள் அல்லது https://www.gmail.com/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை help at sltnet dot lk க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot gmail dot com slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 39.67 ஆகும், இதனை 9/9/2008 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது முப்பத்தொன்பது ரூபாய் அறுபத்தேழு சதம் ஆகும், இதனை ஒன்பது செப்டம்பர் இரண்டாயிரத்து எட்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை luca@outlook.sa க்கு அனுப்புங்கள் அல்லது https://gmail.com ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை luca at outlook dot sa க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash gmail dot com ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை support@mail.fr க்கு அனுப்புங்கள் அல்லது www.example.com/login ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை support at mail dot fr க்கு அனுப்புங்கள் அல்லது w w w dot example dot com slash login ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 186.50 ஆகும், இதனை 14 பிப்ரவரி 1997 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது நூற்று எண்பத்தாறு ரூபாய் ஐம்பது சதம் ஆகும், இதனை பதினான்கு பிப்ரவரி ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஏழு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை jan.devries@yahoo.com க்கு அனுப்புங்கள் அல்லது https://mail.fr/help ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை jan dot devries at yahoo dot com க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash mail dot fr slash help ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை ali35@ziggo.nl க்கு அனுப்புங்கள் அல்லது www.ziggo.nl ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை ali மூன்று ஐந்து at ziggo dot nl க்கு அனுப்புங்கள் அல்லது w w w dot ziggo dot nl ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 8754 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட எண்ணாயிரத்து எழுநூற்று ஐம்பத்து நான்கு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 077-40973471 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஏழு நான்கு பூஜ்ஜியம் ஒன்பது ஏழு மூன்று நான்கு ஏழு ஒன்று என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 43333 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட நாற்பத்து மூன்றாயிரத்து முந்நூற்று முப்பத்து மூன்று என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 449.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு நானூற்று நாற்பத்தொன்பது ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 299 ஆகும், இதனை 6 பிப்ரவரி 2029 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது இருநூற்று தொண்ணூற்று ஒன்பது ரூபாய் ஆகும், இதனை ஆறு பிப்ரவரி இரண்டாயிரத்து இருபத்தொன்பது திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 844975 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட எட்டு லட்சத்து நாற்பத்து நான்காயிரத்து தொள்ளாயிரத்து எழுபத்தைந்து என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 649099 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஆறு லட்சத்து நாற்பத்தொன்பதாயிரத்து தொண்ணூற்று ஒன்பது என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 011 700 3497 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஒன்று ஒன்று ஏழு பூஜ்ஜியம் பூஜ்ஜியம் மூன்று நான்கு ஒன்பது ஏழு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 71 846 2682 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஒன்று எட்டு நான்கு ஆறு இரண்டு ஆறு எட்டு இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.36.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு முப்பத்தாறு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 070-87519575 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு பூஜ்ஜியம் எட்டு ஏழு ஐந்து ஒன்று ஒன்பது ஐந்து ஏழு ஐந்து என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 52854 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஐம்பத்திரண்டாயிரத்து எண்ணூற்று ஐம்பத்து நான்கு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 86 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு எண்பத்தாறு ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 073-95670037 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு மூன்று ஒன்பது ஐந்து ஆறு ஏழு பூஜ்ஜியம் பூஜ்ஜியம் மூன்று ஏழு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை ahmad.b@sltnet.lk க்கு அனுப்புங்கள் அல்லது https://www.yahoo.com.ph/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை ahmad dot b at sltnet dot lk க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot yahoo dot com dot ph slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.773.36 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு எழுநூற்று எழுபத்து மூன்று ரூபாய் முப்பத்தாறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.26,643 ஆகும், இதனை 26 ஏப்ரல் 2013 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது இருபத்தாறாயிரத்து அறுநூற்று நாற்பத்து மூன்று ரூபாய் ஆகும், இதனை இருபத்தாறு ஏப்ரல் இரண்டாயிரத்து பதிமூன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை ahmad.b1@yahoo.com.ph க்கு அனுப்புங்கள் அல்லது www.outlook.sa ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை ahmad dot b ஒன்று at yahoo dot com dot ph க்கு அனுப்புங்கள் அல்லது w w w dot outlook dot sa ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 20882 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட இருபதாயிரத்து எண்ணூற்று எண்பத்திரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 388 ஆகும், இதனை 13 ஜூலை 2020 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது முந்நூற்று எண்பத்தெட்டு ரூபாய் ஆகும், இதனை பதிமூன்று ஜூலை இரண்டாயிரத்து இருபது திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 89.80 ஆகும், இதனை 21/7/2023 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எண்பத்தொன்பது ரூபாய் எண்பது சதம் ஆகும், இதனை இருபத்தொன்று ஜூலை இரண்டாயிரத்து இருபத்து மூன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 078-31455526 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு எட்டு மூன்று ஒன்று நான்கு ஐந்து ஐந்து ஐந்து இரண்டு ஆறு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 70 203 2764 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு பூஜ்ஜியம் இரண்டு பூஜ்ஜியம் மூன்று இரண்டு ஏழு ஆறு நான்கு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 071-30412029 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஒன்று மூன்று பூஜ்ஜியம் நான்கு ஒன்று இரண்டு பூஜ்ஜியம் இரண்டு ஒன்பது என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 77 347 8133 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஏழு மூன்று நான்கு ஏழு எட்டு ஒன்று மூன்று மூன்று என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 011 799 9538 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஒன்று ஒன்று ஏழு ஒன்பது ஒன்பது ஒன்பது ஐந்து மூன்று எட்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 076 779 5783 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஆறு ஏழு ஏழு ஒன்பது ஐந்து ஏழு எட்டு மூன்று என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 079-24377252 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஒன்பது இரண்டு நான்கு மூன்று ஏழு ஏழு இரண்டு ஐந்து இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 729941 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஏழு லட்சத்து இருபத்தொன்பதாயிரத்து தொள்ளாயிரத்து நாற்பத்தொன்று என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 92029 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட தொண்ணூற்று இரண்டாயிரத்து இருபத்தொன்பது என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 078 372 7833 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு எட்டு மூன்று ஏழு இரண்டு ஏழு எட்டு மூன்று மூன்று என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 70 517 3120 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு பூஜ்ஜியம் ஐந்து ஒன்று ஏழு மூன்று ஒன்று இரண்டு பூஜ்ஜியம் என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை nurul_h@yahoo.com க்கு அனுப்புங்கள் அல்லது https://www.sltnet.lk/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை nurul underscore h at yahoo dot com க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot sltnet dot lk slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது LKR 990 ஆகும், இதனை 1 ஆகஸ்ட் 2028 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது தொள்ளாயிரத்து தொண்ணூறு ரூபாய் ஆகும், இதனை ஒன்று ஆகஸ்ட் இரண்டாயிரத்து இருபத்தெட்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது LKR 94 ஆகும், இதனை 2000-01-11 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது தொண்ணூற்று நான்கு ரூபாய் ஆகும், இதனை பதினொன்று ஜனவரி இரண்டாயிரம் திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 15,358.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு பதினைந்தாயிரத்து முந்நூற்று ஐம்பத்தெட்டு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை info72@sapo.pt க்கு அனுப்புங்கள் அல்லது https://web.de ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை info ஏழு இரண்டு at sapo dot pt க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash web dot de ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 875199 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட எட்டு லட்சத்து எழுபத்தைந்தாயிரத்து நூற்று தொண்ணூற்று ஒன்பது என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 775795 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஏழு லட்சத்து எழுபத்தைந்தாயிரத்து எழுநூற்று தொண்ணூற்று ஐந்து என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை support@company.co க்கு அனுப்புங்கள் அல்லது https://sltnet.lk ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை support at company dot co க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash sltnet dot lk ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை fatima@posta.it க்கு அனுப்புங்கள் அல்லது https://gmail.com/help ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை fatima at posta dot it க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash gmail dot com slash help ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 671261 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஆறு லட்சத்து எழுபத்தொன்றாயிரத்து இருநூற்று அறுபத்தொன்று என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை kumar@sapo.pt க்கு அனுப்புங்கள் அல்லது www.yahoo.com.ph/login ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை kumar at sapo dot pt க்கு அனுப்புங்கள் அல்லது w w w dot yahoo dot com dot ph slash login ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை help@sltnet.lk க்கு அனுப்புங்கள் அல்லது https://gmail.com/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை help at sltnet dot lk க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash gmail dot com slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 6647 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஆறாயிரத்து அறுநூற்று நாற்பத்தேழு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை jan.devries64@scicom.com.my க்கு அனுப்புங்கள் அல்லது www.yahoo.com/login ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை jan dot devries ஆறு நான்கு at scicom dot com dot my க்கு அனுப்புங்கள் அல்லது w w w dot yahoo dot com slash login ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது LKR 81,284.90 ஆகும், இதனை 24/2/2020 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எண்பத்தொன்றாயிரத்து இருநூற்று எண்பத்து நான்கு ரூபாய் தொண்ணூறு சதம் ஆகும், இதனை இருபத்து நான்கு பிப்ரவரி இரண்டாயிரத்து இருபது திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 26,563 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு இருபத்தாறாயிரத்து ஐந்நூற்று அறுபத்து மூன்று ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை kasia@sltnet.lk க்கு அனுப்புங்கள் அல்லது https://scicom.com.my/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை kasia at sltnet dot lk க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash scicom dot com dot my slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை ali@example.com க்கு அனுப்புங்கள் அல்லது www.company.co/login ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை ali at example dot com க்கு அனுப்புங்கள் அல்லது w w w dot company dot co slash login ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 9,622.11 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ஒன்பதாயிரத்து அறுநூற்று இருபத்திரண்டு ரூபாய் பதினொன்று சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 4.23 ஆகும், இதனை 5/11/2017 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது நான்கு ரூபாய் இருபத்து மூன்று சதம் ஆகும், இதனை ஐந்து நவம்பர் இரண்டாயிரத்து பதினேழு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 0107 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட பூஜ்ஜியம் ஒன்று பூஜ்ஜியம் ஏழு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை nurul_h@onet.pl க்கு அனுப்புங்கள் அல்லது https://gmail.com/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை nurul underscore h at onet dot pl க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash gmail dot com slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 990.18 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு தொள்ளாயிரத்து தொண்ணூறு ரூபாய் பதினெட்டு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 011 904 0232 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஒன்று ஒன்று ஒன்பது பூஜ்ஜியம் நான்கு பூஜ்ஜியம் இரண்டு மூன்று இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 815.33 ஆகும், இதனை 2015-09-13 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எண்ணூற்று பதினைந்து ரூபாய் முப்பத்து மூன்று சதம் ஆகும், இதனை பதிமூன்று செப்டம்பர் இரண்டாயிரத்து பதினைந்து திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 25,150.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு இருபத்தைந்தாயிரத்து நூற்று ஐம்பது ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 239 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு இருநூற்று முப்பத்தொன்பது ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 45325 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட நாற்பத்தைந்தாயிரத்து முந்நூற்று இருபத்தைந்து என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 508791 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஐந்து லட்சத்து எண்ணாயிரத்து எழுநூற்று தொண்ணூற்று ஒன்று என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 62,000 ஆகும், இதனை 27/12/2032 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது அறுபத்திரண்டாயிரம் ரூபாய் ஆகும், இதனை இருபத்தேழு டிசம்பர் இரண்டாயிரத்து முப்பத்திரண்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 15 ஆகும், இதனை 5 மே 2012 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது பதினைந்து ரூபாய் ஆகும், இதனை ஐந்து மே இரண்டாயிரத்து பன்னிரண்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 79 810 4447 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஒன்பது எட்டு ஒன்று பூஜ்ஜியம் நான்கு நான்கு நான்கு ஏழு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 7155 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஏழாயிரத்து நூற்று ஐம்பத்தைந்து என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 79 803 5992 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஒன்பது எட்டு பூஜ்ஜியம் மூன்று ஐந்து ஒன்பது ஒன்பது இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.19.39 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு பத்தொன்பது ரூபாய் முப்பத்தொன்பது சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 107.73 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு நூற்று ஏழு ரூபாய் எழுபத்து மூன்று சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 89.93 ஆகும், இதனை 2021-06-15 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எண்பத்தொன்பது ரூபாய் தொண்ணூற்று மூன்று சதம் ஆகும், இதனை பதினைந்து ஜூன் இரண்டாயிரத்து இருபத்தொன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 695.50 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு அறுநூற்று தொண்ணூற்று ஐந்து ரூபாய் ஐம்பது சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 23928 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட இருபத்து மூன்றாயிரத்து தொள்ளாயிரத்து இருபத்தெட்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 0510 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட பூஜ்ஜியம் ஐந்து ஒன்று பூஜ்ஜியம் என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.55 ஆகும், இதனை 2 ஜனவரி 2031 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ஐம்பத்தைந்து ரூபாய் ஆகும், இதனை இரண்டு ஜனவரி இரண்டாயிரத்து முப்பத்தொன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 50,190.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ஐம்பதாயிரத்து நூற்று தொண்ணூறு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை jean.dupont79@scicom.com.my க்கு அனுப்புங்கள் அல்லது https://outlook.sa/help ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை jean dot dupont ஏழு ஒன்பது at scicom dot com dot my க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash outlook dot sa slash help ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 18016 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட பதினெட்டாயிரத்து பதினாறு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 41.70 ஆகும், இதனை 15/8/2021 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது நாற்பத்தொன்று ரூபாய் எழுபது சதம் ஆகும், இதனை பதினைந்து ஆகஸ்ட் இரண்டாயிரத்து இருபத்தொன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 60013 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட அறுபதாயிரத்து பதிமூன்று என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.28,995 ஆகும், இதனை 1996-10-24 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது இருபத்தெட்டாயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஐந்து ரூபாய் ஆகும், இதனை இருபத்து நான்கு அக்டோபர் ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஆறு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 89452 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட எண்பத்தொன்பதாயிரத்து நானூற்று ஐம்பத்திரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 3404 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட மூவாயிரத்து நானூற்று நான்கு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.16.48 ஆகும், இதனை 18/1/2012 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது பதினாறு ரூபாய் நாற்பத்தெட்டு சதம் ஆகும், இதனை பதினெட்டு ஜனவரி இரண்டாயிரத்து பன்னிரண்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 079 832 7212 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஒன்பது எட்டு மூன்று இரண்டு ஏழு இரண்டு ஒன்று இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 2112 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட இரண்டாயிரத்து நூற்று பன்னிரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 77 900 6790 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஏழு ஒன்பது பூஜ்ஜியம் பூஜ்ஜியம் ஆறு ஏழு ஒன்பது பூஜ்ஜியம் என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 073 147 7205 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு மூன்று ஒன்று நான்கு ஏழு ஏழு இரண்டு பூஜ்ஜியம் ஐந்து என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 66 ஆகும், இதனை டிசம்பர் 24 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது அறுபத்தாறு ரூபாய் ஆகும், இதனை டிசம்பர் இருபத்து நான்கு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 35 ஆகும், இதனை ஜூன் 25 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது முப்பத்தைந்து ரூபாய் ஆகும், இதனை ஜூன் இருபத்தைந்து திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.45,082.05 ஆகும், இதனை 26 டிசம்பர் 2026 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது நாற்பத்தைந்தாயிரத்து எண்பத்திரண்டு ரூபாய் ஐந்து சதம் ஆகும், இதனை இருபத்தாறு டிசம்பர் இரண்டாயிரத்து இருபத்தாறு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 94975 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட தொண்ணூற்று நான்காயிரத்து தொள்ளாயிரத்து எழுபத்தைந்து என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை info5@correo.es க்கு அனுப்புங்கள் அல்லது https://yahoo.com.ph ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை info ஐந்து at correo dot es க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash yahoo dot com dot ph ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 890 ஆகும், இதனை 9/4/2032 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எண்ணூற்று தொண்ணூறு ரூபாய் ஆகும், இதனை ஒன்பது ஏப்ரல் இரண்டாயிரத்து முப்பத்திரண்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை maria60@onet.pl க்கு அனுப்புங்கள் அல்லது https://www.yahoo.com/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை maria ஆறு பூஜ்ஜியம் at onet dot pl க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot yahoo dot com slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 54052 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஐம்பத்து நான்காயிரத்து ஐம்பத்திரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.7 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ஏழு ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை mei.ling@mail.fr க்கு அனுப்புங்கள் அல்லது https://www.gmail.co.id/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை mei dot ling at mail dot fr க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot gmail dot co dot id slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 79 289 8272 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு ஒன்பது இரண்டு எட்டு ஒன்பது எட்டு இரண்டு ஏழு இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.26,077 ஆகும், இதனை 3 ஜூலை 2021 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது இருபத்தாறாயிரத்து எழுபத்தேழு ரூபாய் ஆகும், இதனை மூன்று ஜூலை இரண்டாயிரத்து இருபத்தொன்று திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை kumar71@mail.fr க்கு அனுப்புங்கள் அல்லது https://onet.pl/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை kumar ஏழு ஒன்று at mail dot fr க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash onet dot pl slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது LKR 95,170.50 ஆகும், இதனை 16/3/1996 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது தொண்ணூற்று ஐந்தாயிரத்து நூற்று எழுபது ரூபாய் ஐம்பது சதம் ஆகும், இதனை பதினாறு மார்ச் ஆயிரத்து தொள்ளாயிரத்து தொண்ணூற்று ஆறு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 992.74 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு தொள்ளாயிரத்து தொண்ணூற்று இரண்டு ரூபாய் எழுபத்து நான்கு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 1150 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஆயிரத்து நூற்று ஐம்பது என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை help@correo.es க்கு அனுப்புங்கள் அல்லது https://mail.fr/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை help at correo dot es க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash mail dot fr slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 070 260 6726 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு பூஜ்ஜியம் இரண்டு ஆறு பூஜ்ஜியம் ஆறு ஏழு இரண்டு ஆறு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 557972 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட ஐந்து லட்சத்து ஐம்பத்தேழாயிரத்து தொள்ளாயிரத்து எழுபத்திரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை info37@scicom.com.my க்கு அனுப்புங்கள் அல்லது www.gmail.co.id/login ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை info மூன்று ஏழு at scicom dot com dot my க்கு அனுப்புங்கள் அல்லது w w w dot gmail dot co dot id slash login ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 46882 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட நாற்பத்தாறாயிரத்து எண்ணூற்று எண்பத்திரண்டு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.54,638 ஆகும், இதனை 5/2/2005 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ஐம்பத்து நான்காயிரத்து அறுநூற்று முப்பத்தெட்டு ரூபாய் ஆகும், இதனை ஐந்து பிப்ரவரி இரண்டாயிரத்து ஐந்து திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை omar43@yahoo.com க்கு அனுப்புங்கள் அல்லது www.gmail.com ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை omar நான்கு மூன்று at yahoo dot com க்கு அனுப்புங்கள் அல்லது w w w dot gmail dot com ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs.799.90 ஆகும், இதனை 2019-07-28 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது எழுநூற்று தொண்ணூற்று ஒன்பது ரூபாய் தொண்ணூறு சதம் ஆகும், இதனை இருபத்தெட்டு ஜூலை இரண்டாயிரத்து பத்தொன்பது திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.93,622.38 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு தொண்ணூற்று மூன்றாயிரத்து அறுநூற்று இருபத்திரண்டு ரூபாய் முப்பத்தெட்டு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை mei.ling48@gmail.com க்கு அனுப்புங்கள் அல்லது https://outlook.sa ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை mei dot ling நான்கு எட்டு at gmail dot com க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash outlook dot sa ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 896254 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட எட்டு லட்சத்து தொண்ணூற்று ஆறாயிரத்து இருநூற்று ஐம்பத்து நான்கு என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட 088479 என்ற குறியீட்டை உள்ளிடவும்."}, {"role": "assistant", "content": "உறுதிப்படுத்தலுக்காக உங்கள் மொபைலுக்கு அனுப்பப்பட்ட பூஜ்ஜியம் எட்டு எட்டு நான்கு ஏழு ஒன்பது என்ற குறியீட்டை உள்ளிடவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 077 521 7408 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஏழு ஐந்து இரண்டு ஒன்று ஏழு நான்கு பூஜ்ஜியம் எட்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 32.50 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு முப்பத்திரண்டு ரூபாய் ஐம்பது சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது ரூ. 614.90 ஆகும், இதனை 13/8/2032 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது அறுநூற்று பதினான்கு ரூபாய் தொண்ணூறு சதம் ஆகும், இதனை பதிமூன்று ஆகஸ்ட் இரண்டாயிரத்து முப்பத்திரண்டு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.691.50 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு அறுநூற்று தொண்ணூற்று ஒன்று ரூபாய் ஐம்பது சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 071 783 3820 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஒன்று ஏழு எட்டு மூன்று மூன்று எட்டு இரண்டு பூஜ்ஜியம் என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய +94 74 995 2232 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பிளஸ் ஒன்பது நான்கு ஏழு நான்கு ஒன்பது ஒன்பது ஐந்து இரண்டு இரண்டு மூன்று இரண்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை info@bank.my க்கு அனுப்புங்கள் அல்லது https://ziggo.nl ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை info at bank dot my க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash ziggo dot nl ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 218.50 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு இருநூற்று பதினெட்டு ரூபாய் ஐம்பது சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 076-21966836 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஆறு இரண்டு ஒன்று ஒன்பது ஆறு ஆறு எட்டு மூன்று ஆறு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை maria@bank.my க்கு அனுப்புங்கள் அல்லது https://correo.es/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை maria at bank dot my க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash correo dot es slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs.97.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு தொண்ணூற்று ஏழு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 78,276 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு எழுபத்தெட்டாயிரத்து இருநூற்று எழுபத்தாறு ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 077 003 6615 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஏழு பூஜ்ஜியம் பூஜ்ஜியம் மூன்று ஆறு ஆறு ஒன்று ஐந்து என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 818.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு எண்ணூற்று பதினெட்டு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 076-93360668 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஆறு ஒன்பது மூன்று மூன்று ஆறு பூஜ்ஜியம் ஆறு ஆறு எட்டு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய புகாரை பதிவு செய்ய 079 231 4836 என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}, {"role": "assistant", "content": "உங்களுடைய புகாரை பதிவு செய்ய பூஜ்ஜியம் ஏழு ஒன்பது இரண்டு மூன்று ஒன்று நான்கு எட்டு மூன்று ஆறு என்ற இலக்கத்திற்கு அழைப்பு விடுக்கவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு Rs. 428.90 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு நானூற்று இருபத்தெட்டு ரூபாய் தொண்ணூறு சதம் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை dilani@gmail.co.id க்கு அனுப்புங்கள் அல்லது https://mail.fr/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை dilani at gmail dot co dot id க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash mail dot fr slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது Rs. 62,717 ஆகும், இதனை 27 பிப்ரவரி 2014 திகதிக்கு முன் செலுத்தவும்."}, {"role": "assistant", "content": "உங்களுடைய மின்சாரக் கட்டணக் கொடுப்பனவானது அறுபத்திரண்டாயிரத்து எழுநூற்று பதினேழு ரூபாய் ஆகும், இதனை இருபத்தேழு பிப்ரவரி இரண்டாயிரத்து பதினான்கு திகதிக்கு முன் செலுத்தவும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை fatima@scicom.com.my க்கு அனுப்புங்கள் அல்லது https://gmail.com/faq ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை fatima at scicom dot com dot my க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash gmail dot com slash faq ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு ரூ. 99 டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}, {"role": "assistant", "content": "கொழும்புவில் இருந்து யாழ்ப்பாணத்திற்கு அனுப்பப்படும் பார்சல்களுக்கு தொண்ணூற்று ஒன்பது ரூபாய் டெலிவரி கட்டணமாக வசூலிக்கப்படும்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "படிவத்தை billing13@scicom.com.my க்கு அனுப்புங்கள் அல்லது https://www.ziggo.nl/support ஐ பார்வையிடுங்கள்."}, {"role": "assistant", "content": "படிவத்தை billing ஒன்று மூன்று at scicom dot com dot my க்கு அனுப்புங்கள் அல்லது h t t p s colon slash slash www dot ziggo dot nl slash support ஐ பார்வையிடுங்கள்."}], "lang": "ta-LK", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත REF-2024-0113 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත REF දෙක බිංදුව දෙක හතර බිංදුව එක එක තුන ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය QZ8501 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය QZ අට පහ බිංදුව එක වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත X-12340567 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත X එක දෙක තුන හතර බිංදුව පහ හය හත ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය INV0472 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය INV බිංදුව හතර හත දෙක වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත B2-14 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත B දෙක එක හතර ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය X-12340567 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය X එක දෙක තුන හතර බිංදුව පහ හය හත වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 5763 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ හත හය තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය MH370 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය MH තුන හත බිංදුව වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය ABC1234 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය ABC එක දෙක තුන හතර වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 929033 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය දෙක නවය බිංදුව තුන තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත ABC1234 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත ABC එක දෙක තුන හතර ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත PO 55219 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත PO පහ පහ දෙක එක නවය ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත INV0472 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත INV බිංදුව හතර හත දෙක ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය REF-2024-0113 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය REF දෙක බිංදුව දෙක හතර බිංදුව එක එක තුන වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය WXY 1234 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය WXY එක දෙක තුන හතර වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය B2-14 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය B දෙක එක හතර වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 1024 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක බිංදුව දෙක හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය A12 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය A එක දෙක වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත AK6120 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත AK හය එක දෙක බිංදුව ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 93927 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය තුන නවය දෙක හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත TK-88213 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත TK අට අට දෙක එක තුන ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 162138 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක හය දෙක එක තුන අට කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 9231 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය දෙක තුන එක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 1295 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක දෙක නවය පහ කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 99787 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය නවය හත අට හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 79286 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත නවය දෙක අට හය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත A12 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත A එක දෙක ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 894688 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට අට නවය හතර හය අට අට කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 957803 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය පහ හත අට බිංදුව තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 28477 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක අට හතර හත හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 7337 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත තුන තුන හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය PO 55219 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය PO පහ පහ දෙක එක නවය වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 195516 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක නවය පහ පහ එක හය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය AK6120 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය AK හය එක දෙක බිංදුව වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 4323 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හතර තුන දෙක තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 57977 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ හත නවය හත හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 2243 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක දෙක හතර තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 890316 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට අට නවය බිංදුව තුන එක හය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 4620 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හතර හය දෙක බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 10086 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක බිංදුව බිංදුව අට හය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 91167 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය එක එක හය හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත MH370 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත MH තුන හත බිංදුව ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 51023 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ එක බිංදුව දෙක තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 88450 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට අට අට හතර පහ බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත WXY 1234 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත WXY එක දෙක තුන හතර ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 12162 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට එක දෙක එක හය දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 577864 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ හත හත අට හය හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 907328 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය බිංදුව හත තුන දෙක අට කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 078854 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට බිංදුව හත අට අට පහ හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 2094 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක බිංදුව නවය හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 84977 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට අට හතර නවය හත හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 0717 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට බිංදුව හත එක හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 200111 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක බිංදුව බිංදුව එක එක එක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 588445 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ අට අට හතර හතර පහ කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 262087 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක හය දෙක බිංදුව අට හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 3822 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට තුන අට දෙක දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 23412 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක තුන හතර එක දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 253959 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක පහ තුන නවය පහ නවය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 30034 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට තුන බිංදුව බිංදුව තුන හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 859421 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට අට පහ නවය හතර දෙක එක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 0256 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට බිංදුව දෙක පහ හය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 6352 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හය තුන පහ දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 6285 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හය දෙක අට පහ කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 717779 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත එක හත හත හත නවය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 9132 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට නවය එක තුන දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 4062 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හතර බිංදුව හය දෙක කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 42060 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හතර දෙක බිංදුව හය බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 20847 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක බිංදුව අට හතර හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 6177 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හය එක හත හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 202180 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක බිංදුව දෙක එක අට බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 79397 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත නවය තුන නවය හත කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 230690 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට දෙක තුන බිංදුව හය නවය බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 598018 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ නවය අට බිංදුව එක අට කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 534784 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට පහ තුන හතර හත අට හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 761903 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත හය එක නවය බිංදුව තුන කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුම් හැඳුනුම්පත QZ8501 ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුම් හැඳුනුම්පත QZ අට පහ බිංදුව එක ලෙස සඳහන් කර පණිවිඩයක් එවන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 67910 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හය හත නවය එක බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 01750 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට බිංදුව එක හත පහ බිංදුව කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 74639 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හත හතර හය තුන නවය කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගනුදෙනු අංකය TK-88213 වේ."}, {"role": "assistant", "content": "ඔබේ ගනුදෙනු අංකය TK අට අට දෙක එක තුන වේ."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට ඇතුළු වීමට 638054 කේතය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමට ඇතුළු වීමට හය තුන අට බිංදුව පහ හතර කේතය භාවිතා කරන්න."}], "lang": "si", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 77.6 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng pitumpu't pito punto anim na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.mail.fr."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot mail dot fr."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 41.80 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng apatnapu't isa punto walo zerong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.sltnet.lk/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot sltnet dot lk slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 41.68 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng apatnapu't isa punto anim walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.web.de/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot web dot de slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 26,459.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawampu't anim na libo apat na raan at limampu't siyam na piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-3 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikatlo na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://web.de."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa web dot de."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1963 sa loob ng 13-20."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at animnapu't tatlo sa loob ng labintatlo hanggang dalawampu."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-24 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikadalawampu't apat na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-12 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabindalawa na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na P93,624.11 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na siyamnapu't tatlong libo anim na raan at dalawampu't apat na piso at labing-isang sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 89.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na walumpu't siyam na piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 99.7 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng siyamnapu't siyam punto pitong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.yahoo.com.ph/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot yahoo dot com dot ph slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 78.26 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng pitumpu't walo punto dalawa anim na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 37.31 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng tatlumpu't pito punto tatlo isang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.bank.my/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot bank dot my slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.gmail.com/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot gmail dot com slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.sltnet.lk."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot sltnet dot lk."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2030 sa loob ng 10 - 13."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at tatlumpu sa loob ng sampu hanggang labintatlo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2023 sa loob ng 2–9."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawampu't tatlo sa loob ng dalawa hanggang siyam."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 18.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na labingwalong piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 57.8 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng limampu't pito punto walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 680 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na anim na raan at walumpung piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 80.4 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng walumpu punto apat na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 22.5 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng dalawampu't dalawa punto limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 28.67 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng dalawampu't walo punto anim pitong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱15.57 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na labinlimang piso at limampu't pitong sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱27 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawampu't pitong piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 41.9 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng apatnapu't isa punto siyam na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 93.66 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng siyamnapu't tatlo punto anim anim na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-2 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalawa na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 11.5 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng labing-isa punto limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.sapo.pt/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot sapo dot pt slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱250.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawang daan at limampung piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.web.de."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot web dot de."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 58.6 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng limampu't walo punto anim na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-17 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabimpito na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 778 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na pitong daan at pitumpu't walong piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-7 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikapito na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na Php 838.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na walong daan at tatlumpu't walong piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱216 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawang daan at labing-anim na piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 52.8 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng limampu't dalawa punto walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 55.34 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng limampu't lima punto tatlo apat na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-6 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikaanim na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2034 sa loob ng 8-14."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at tatlumpu't apat sa loob ng walo hanggang labing-apat."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.mail.fr/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot mail dot fr slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱22 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawampu't dalawang piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 94.08 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng siyamnapu't apat punto zero walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://correo.es/help."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa correo dot es slash help."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://sltnet.lk."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa sltnet dot lk."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1994 sa loob ng 19-23."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at siyamnapu't apat sa loob ng labinsiyam hanggang dalawampu't tatlo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na P101.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na isang daan at isang piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.outlook.sa/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot outlook dot sa slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1998 sa loob ng 29-37."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at siyamnapu't walo sa loob ng dalawampu't siyam hanggang tatlumpu't pito."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 23.62 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng dalawampu't tatlo punto anim dalawang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1959 sa loob ng 29-39."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at limampu't siyam sa loob ng dalawampu't siyam hanggang tatlumpu't siyam."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 84.5 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng walumpu't apat punto limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.sltnet.lk/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot sltnet dot lk slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-19 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabinsiyam na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.scicom.com.my."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot scicom dot com dot my."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 9.85 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng siyam punto walo limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2011 sa loob ng 1 - 5."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at labing-isa sa loob ng isa hanggang lima."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 90,896.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na siyamnapung libo walong daan at siyamnapu't anim na piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://company.co/help."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa company dot co slash help."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 73.42 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng pitumpu't tatlo punto apat dalawang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://bank.my."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa bank dot my."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-27 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikadalawampu't pito na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2009 sa loob ng 6-16."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at siyam sa loob ng anim hanggang labing-anim."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1981 sa loob ng 18 - 22."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at walumpu't isa sa loob ng labingwalo hanggang dalawampu't dalawa."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1966 sa loob ng 28–31."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at animnapu't anim sa loob ng dalawampu't walo hanggang tatlumpu't isa."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 18.91 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng labingwalo punto siyam isang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://onet.pl/faq."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa onet dot pl slash faq."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1951 sa loob ng 15-22."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at limampu't isa sa loob ng labinlima hanggang dalawampu't dalawa."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-20 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikadalawampu na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2008 sa loob ng 29–33."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at walo sa loob ng dalawampu't siyam hanggang tatlumpu't tatlo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1989 sa loob ng 24 - 28."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at walumpu't siyam sa loob ng dalawampu't apat hanggang dalawampu't walo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2011 sa loob ng 25-28."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at labing-isa sa loob ng dalawampu't lima hanggang dalawampu't walo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2025 sa loob ng 21-29."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawampu't lima sa loob ng dalawampu't isa hanggang dalawampu't siyam."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2021 sa loob ng 27 - 30."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawampu't isa sa loob ng dalawampu't pito hanggang tatlumpu."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-14 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabing-apat na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1997 sa loob ng 22-26."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at siyamnapu't pito sa loob ng dalawampu't dalawa hanggang dalawampu't anim."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na Php 77,623 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na pitumpu't pitong libo anim na raan at dalawampu't tatlong piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1954 sa loob ng 30-34."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at limampu't apat sa loob ng tatlumpu hanggang tatlumpu't apat."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 9.90 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na siyam na piso at siyamnapung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2033 sa loob ng 24-34."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at tatlumpu't tatlo sa loob ng dalawampu't apat hanggang tatlumpu't apat."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na Php 74.59 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na pitumpu't apat na piso at limampu't siyam na sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-31 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikatatlumpu't isa na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 63.58 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng animnapu't tatlo punto lima walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1951 sa loob ng 11 - 17."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at limampu't isa sa loob ng labing-isa hanggang labimpito."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-18 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabingwalo na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2008 sa loob ng 26-29."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at walo sa loob ng dalawampu't anim hanggang dalawampu't siyam."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na Php 18.90 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na labingwalong piso at siyamnapung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 86.1 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng walumpu't anim punto isang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-16 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong ikalabing-anim na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://posta.it."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa posta dot it."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2032 sa loob ng 8-16."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at tatlumpu't dalawa sa loob ng walo hanggang labing-anim."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 84,242.32 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na walumpu't apat na libo dalawang daan at apatnapu't dalawang piso at tatlumpu't dalawang sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mayroon kaming special offer para sa iyong ika-1 na buwan bilang subscriber."}, {"role": "assistant", "content": "Mayroon kaming special offer para sa iyong una na buwan bilang subscriber."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 63.86 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng animnapu't tatlo punto walo anim na percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 91.75 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng siyamnapu't isa punto pito limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 61.8 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng animnapu't isa punto walong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 61.51 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng animnapu't isa punto lima isang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2002 sa loob ng 13–18."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawa sa loob ng labintatlo hanggang labingwalo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://yahoo.com.ph/faq."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa yahoo dot com dot ph slash faq."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1991 sa loob ng 8-12."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at siyamnapu't isa sa loob ng walo hanggang labindalawa."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2010 sa loob ng 1-3."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at sampu sa loob ng isa hanggang tatlo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1975 sa loob ng 24–34."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at pitumpu't lima sa loob ng dalawampu't apat hanggang tatlumpu't apat."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1983 sa loob ng 14-19."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at walumpu't tatlo sa loob ng labing-apat hanggang labinsiyam."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://bank.my/help."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa bank dot my slash help."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.yahoo.com/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot yahoo dot com slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://www.yahoo.com.ph/support."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot yahoo dot com dot ph slash support."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.onet.pl/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot onet dot pl slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://yahoo.com/faq."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa yahoo dot com slash faq."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://sltnet.lk/faq."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa sltnet dot lk slash faq."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2029 sa loob ng 17 - 27."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawampu't siyam sa loob ng labimpito hanggang dalawampu't pito."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://yahoo.com/help."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa yahoo dot com slash help."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na P2,549 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawang libo limang daan at apatnapu't siyam na piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱23.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na dalawampu't tatlong piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://onet.pl."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa onet dot pl."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1975 sa loob ng 15–17."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at pitumpu't lima sa loob ng labinlima hanggang labimpito."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na PHP 13,928 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na labintatlong libo siyam na raan at dalawampu't walong piso kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 19.85 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng labinsiyam punto walo limang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na ₱390.50 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na tatlong daan at siyamnapung piso at limampung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa https://sapo.pt/help."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa sapo dot pt slash help."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 36.3 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng tatlumpu't anim punto tatlong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 2002 sa loob ng 1-8."}, {"role": "assistant", "content": "Ang event ay gaganapin sa dalawang libo at dalawa sa loob ng isa hanggang walo."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Maaari niyo pong i-update ang inyong profile sa www.gmail.com/login."}, {"role": "assistant", "content": "Maaari niyo pong i-update ang inyong profile sa w w w dot gmail dot com slash login."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang event ay gaganapin sa 1998 sa loob ng 14 - 15."}, {"role": "assistant", "content": "Ang event ay gaganapin sa isang libo siyam na raan at siyamnapu't walo sa loob ng labing-apat hanggang labinlima."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 7.1 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng pito punto isang percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong account ay may limit na Php 337.90 kada araw."}, {"role": "assistant", "content": "Ang iyong account ay may limit na tatlong daan at tatlumpu't pitong piso at siyamnapung sentimo kada araw."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang shipping fee ay tumaas ng 2.3 percent dahil sa fuel surcharge."}, {"role": "assistant", "content": "Ang shipping fee ay tumaas ng dalawa punto tatlong percent dahil sa fuel surcharge."}], "lang": "tl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://example.com/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: example نقطة com شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://correo.es/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: correo نقطة es شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى dilani@gmail.co.id، المجموع النهائي هو ١٠٢.٩٠ دولار."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى dilani آت gmail نقطة co نقطة id، المجموع النهائي هو مائة واثنان دولاراً وتسعون سنتاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو WXY 1234 وسنقوم بمعالجته خلال 131 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو WXY واحد اثنان ثلاثة أربعة وسنقوم بمعالجته خلال مائة وواحد وثلاثون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي WXY ١٢٣٤ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي WXY واحد اثنان ثلاثة أربعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى luca@sltnet.lk، المجموع النهائي هو ٢,٧٢٢ ر.س."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى luca آت sltnet نقطة lk، المجموع النهائي هو ألفان وسبعمائة واثنان وعشرون ريالاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى ali@gmail.com، المجموع النهائي هو ٦٨,٨٣٣.٩٠ ريال."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى ali آت gmail نقطة com، المجموع النهائي هو ثمانية وستون ألفاً وثمانمائة وثلاثة وثلاثون ريالاً وتسعون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى info96@web.de، المجموع النهائي هو 46,550 درهم."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى info تسعة ستة آت web نقطة de، المجموع النهائي هو ستة وأربعون ألفاً وخمسمائة وخمسون درهماً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي REF-2024-0113 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي REF اثنان صفر اثنان أربعة صفر واحد واحد ثلاثة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://company.co/help"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: company نقطة co شرطة مائلة help"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean.dupont@sapo.pt، المجموع النهائي هو 57.90 دولار."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean نقطة dupont آت sapo نقطة pt، المجموع النهائي هو سبعة وخمسون دولاراً وتسعون سنتاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean.dupont@posta.it، المجموع النهائي هو 25,091.50 ر.س."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean نقطة dupont آت posta نقطة it، المجموع النهائي هو خمسة وعشرون ألفاً وواحد وتسعون ريالاً وخمسون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو PO 55219 وسنقوم بمعالجته خلال 6 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو PO خمسة خمسة اثنان واحد تسعة وسنقوم بمعالجته خلال ستة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى fatima@company.co، المجموع النهائي هو ٩٦.٩٠ ر.س."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى fatima آت company نقطة co، المجموع النهائي هو ستة وتسعون ريالاً وتسعون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى siti@yahoo.com.ph، المجموع النهائي هو 73,678.50 ريال."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى siti آت yahoo نقطة com نقطة ph، المجموع النهائي هو ثلاثة وسبعون ألفاً وستمائة وثمانية وسبعون ريالاً وخمسون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia@scicom.com.my، المجموع النهائي هو AED ٩٤٠.٥٢."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia آت scicom نقطة com نقطة my، المجموع النهائي هو تسعمائة وأربعون درهماً واثنان وخمسون فلساً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 043401 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق صفر أربعة ثلاثة أربعة صفر واحد لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى info23@correo.es، المجموع النهائي هو $50."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى info اثنان ثلاثة آت correo نقطة es، المجموع النهائي هو خمسون دولاراً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي WXY 1234 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي WXY واحد اثنان ثلاثة أربعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://sltnet.lk/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: sltnet نقطة lk شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي X-١٢٣٤٠٥٦٧ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي X واحد اثنان ثلاثة أربعة صفر خمسة ستة سبعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو MH370 وسنقوم بمعالجته خلال 39 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو MH ثلاثة سبعة صفر وسنقوم بمعالجته خلال تسعة وثلاثون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي MH370 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي MH ثلاثة سبعة صفر للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي X-12340567 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي X واحد اثنان ثلاثة أربعة صفر خمسة ستة سبعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 5261 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق خمسة اثنان ستة واحد لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://web.de"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: web نقطة de"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى anna@bank.my، المجموع النهائي هو $836."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى anna آت bank نقطة my، المجموع النهائي هو ثمانمائة وستة وثلاثون دولاراً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: www.example.com/login"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة example نقطة com شرطة مائلة login"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي AK6120 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي AK ستة واحد اثنان صفر للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو PO ٥٥٢١٩ وسنقوم بمعالجته خلال ٣٠٧ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو PO خمسة خمسة اثنان واحد تسعة وسنقوم بمعالجته خلال ثلاثمائة وسبعة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو PO 55219 وسنقوم بمعالجته خلال 2 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو PO خمسة خمسة اثنان واحد تسعة وسنقوم بمعالجته خلال اثنان أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي B2-14 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي B اثنان واحد أربعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://yahoo.com/help"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: yahoo نقطة com شرطة مائلة help"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٦٤٤٢ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ستة أربعة أربعة اثنان لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: www.bank.my/login"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة bank نقطة my شرطة مائلة login"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 0451 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق صفر أربعة خمسة واحد لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو QZ8501 وسنقوم بمعالجته خلال 2 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو QZ ثمانية خمسة صفر واحد وسنقوم بمعالجته خلال اثنان أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي TK-88213 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي TK ثمانية ثمانية اثنان واحد ثلاثة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://ziggo.nl/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: ziggo نقطة nl شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jan.devries@web.de، المجموع النهائي هو USD 45."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jan نقطة devries آت web نقطة de، المجموع النهائي هو خمسة وأربعون دولاراً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو QZ٨٥٠١ وسنقوم بمعالجته خلال ١٨٥ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو QZ ثمانية خمسة صفر واحد وسنقوم بمعالجته خلال مائة وخمسة وثمانون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى nurul_h@yahoo.com.ph، المجموع النهائي هو $99,210."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى nurul underscore h آت yahoo نقطة com نقطة ph، المجموع النهائي هو تسعة وتسعون ألفاً ومئتان وعشرة دولارات."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 4884 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق أربعة ثمانية ثمانية أربعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى kumar7@gmail.com، المجموع النهائي هو 33,531 جنيه."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى kumar سبعة آت gmail نقطة com، المجموع النهائي هو ثلاثة وثلاثون ألفاً وخمسمائة وواحد وثلاثون جنيهاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي ABC1234 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي ABC واحد اثنان ثلاثة أربعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو TK-88213 وسنقوم بمعالجته خلال 8 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو TK ثمانية ثمانية اثنان واحد ثلاثة وسنقوم بمعالجته خلال ثمانية أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي INV٠٤٧٢ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي INV صفر أربعة سبعة اثنان للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 95388 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق تسعة خمسة ثلاثة ثمانية ثمانية لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://www.bank.my/support"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة bank نقطة my شرطة مائلة support"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو B٢-١٤ وسنقوم بمعالجته خلال ٥٧٢ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو B اثنان واحد أربعة وسنقوم بمعالجته خلال خمسمائة واثنان وسبعون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: www.outlook.sa/login"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة outlook نقطة sa شرطة مائلة login"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي A12 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي A واحد اثنان للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٤٠٢٩٦ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق أربعة صفر اثنان تسعة ستة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو TK-88213 وسنقوم بمعالجته خلال 57 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو TK ثمانية ثمانية اثنان واحد ثلاثة وسنقوم بمعالجته خلال سبعة وخمسون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://bank.my/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: bank نقطة my شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو MH370 وسنقوم بمعالجته خلال 45 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو MH ثلاثة سبعة صفر وسنقوم بمعالجته خلال خمسة وأربعون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://correo.es"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: correo نقطة es"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean.dupont@sltnet.lk، المجموع النهائي هو ٨٥١.٧٠ دولار."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jean نقطة dupont آت sltnet نقطة lk، المجموع النهائي هو ثمانمائة وواحد وخمسون دولاراً وسبعون سنتاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: www.mail.fr/login"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة mail نقطة fr شرطة مائلة login"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://gmail.com/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: gmail نقطة com شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى fatima77@scicom.com.my، المجموع النهائي هو 674.90 دولار."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى fatima سبعة سبعة آت scicom نقطة com نقطة my، المجموع النهائي هو ستمائة وأربعة وسبعون دولاراً وتسعون سنتاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://www.example.com/support"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة example نقطة com شرطة مائلة support"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 534037 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق خمسة ثلاثة أربعة صفر ثلاثة سبعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى help@example.com، المجموع النهائي هو AED 53,848."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى help آت example نقطة com، المجموع النهائي هو ثلاثة وخمسون ألفاً وثمانمائة وثمانية وأربعون درهماً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو TK-88213 وسنقوم بمعالجته خلال 28 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو TK ثمانية ثمانية اثنان واحد ثلاثة وسنقوم بمعالجته خلال ثمانية وعشرون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jan.devries@example.com، المجموع النهائي هو ٥٢.٢٣ ر.س."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى jan نقطة devries آت example نقطة com، المجموع النهائي هو اثنان وخمسون ريالاً وثلاثة وعشرون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia@gmail.com، المجموع النهائي هو ٦١ ريال."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia آت gmail نقطة com، المجموع النهائي هو واحد وستون ريالاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 1368 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق واحد ثلاثة ستة ثمانية لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 63551 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ستة ثلاثة خمسة خمسة واحد لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 643758 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ستة أربعة ثلاثة سبعة خمسة ثمانية لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 8633 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ثمانية ستة ثلاثة ثلاثة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://gmail.co.id/help"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: gmail نقطة co نقطة id شرطة مائلة help"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي PO 55219 للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي PO خمسة خمسة اثنان واحد تسعة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو X-12340567 وسنقوم بمعالجته خلال 4 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو X واحد اثنان ثلاثة أربعة صفر خمسة ستة سبعة وسنقوم بمعالجته خلال أربعة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو QZ8501 وسنقوم بمعالجته خلال 10 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو QZ ثمانية خمسة صفر واحد وسنقوم بمعالجته خلال عشرة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو MH٣٧٠ وسنقوم بمعالجته خلال ٦٣٠ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو MH ثلاثة سبعة صفر وسنقوم بمعالجته خلال ستمائة وثلاثون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://www.gmail.co.id/support"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة gmail نقطة co نقطة id شرطة مائلة support"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 762695 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق سبعة ستة اثنان ستة تسعة خمسة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو REF-٢٠٢٤-٠١١٣ وسنقوم بمعالجته خلال ٦٠ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو REF اثنان صفر اثنان أربعة صفر واحد واحد ثلاثة وسنقوم بمعالجته خلال ستون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 77667 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق سبعة سبعة ستة ستة سبعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى siti@bank.my، المجموع النهائي هو 939.50 درهم."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى siti آت bank نقطة my، المجموع النهائي هو تسعمائة وتسعة وثلاثون درهماً وخمسون فلساً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٢٤٥٩٣ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق اثنان أربعة خمسة تسعة ثلاثة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي AK٦١٢٠ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي AK ستة واحد اثنان صفر للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى support@mail.fr، المجموع النهائي هو AED 80,870.50."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى support آت mail نقطة fr، المجموع النهائي هو ثمانون ألفاً وثمانمائة وسبعون درهماً وخمسون فلساً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://web.de/help"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: web نقطة de شرطة مائلة help"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٥٨٥٧ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق خمسة ثمانية خمسة سبعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو WXY 1234 وسنقوم بمعالجته خلال 692 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو WXY واحد اثنان ثلاثة أربعة وسنقوم بمعالجته خلال ستمائة واثنان وتسعون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٤٤٦٥ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق أربعة أربعة ستة خمسة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://posta.it"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: posta نقطة it"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٥٩٠١٨٤ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق خمسة تسعة صفر واحد ثمانية أربعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى kasia@mail.fr، المجموع النهائي هو SAR ٩,٦٢٢.٩٠."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى kasia آت mail نقطة fr، المجموع النهائي هو تسعة آلاف وستمائة واثنان وعشرون ريالاً وتسعون هللةً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia@gmail.co.id، المجموع النهائي هو 405.82 د.إ."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى sofia آت gmail نقطة co نقطة id، المجموع النهائي هو أربعمائة وخمسة دراهم واثنان وثمانون فلساً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 523295 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق خمسة اثنان ثلاثة اثنان تسعة خمسة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو QZ٨٥٠١ وسنقوم بمعالجته خلال ٧ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو QZ ثمانية خمسة صفر واحد وسنقوم بمعالجته خلال سبعة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 86479 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ثمانية ستة أربعة سبعة تسعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى maria٦٦@sapo.pt، المجموع النهائي هو ١٨,٠٦٤ درهم."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى maria ستة ستة آت sapo نقطة pt، المجموع النهائي هو ثمانية عشر ألفاً وأربعة وستون درهماً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى dilani@web.de، المجموع النهائي هو USD ٥١٩."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى dilani آت web نقطة de، المجموع النهائي هو خمسمائة وتسعة عشر دولاراً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو B٢-١٤ وسنقوم بمعالجته خلال ٤ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو B اثنان واحد أربعة وسنقوم بمعالجته خلال أربعة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو WXY 1234 وسنقوم بمعالجته خلال 640 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو WXY واحد اثنان ثلاثة أربعة وسنقوم بمعالجته خلال ستمائة وأربعون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://yahoo.com.ph/faq"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: yahoo نقطة com نقطة ph شرطة مائلة faq"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى nurul_h46@posta.it، المجموع النهائي هو EGP 580.90."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى nurul underscore h أربعة ستة آت posta نقطة it، المجموع النهائي هو خمسمائة وثمانون جنيهاً وتسعون قرشاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو MH370 وسنقوم بمعالجته خلال 451 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو MH ثلاثة سبعة صفر وسنقوم بمعالجته خلال أربعمائة وواحد وخمسون أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://ziggo.nl"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: ziggo نقطة nl"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى juan@onet.pl، المجموع النهائي هو ٤٧.٥٠ دولار."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى juan آت onet نقطة pl، المجموع النهائي هو سبعة وأربعون دولاراً وخمسون سنتاً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "لقد تم إرسال تفاصيل الفاتورة إلى ahmad.b@gmail.co.id، المجموع النهائي هو $24,491."}, {"role": "assistant", "content": "لقد تم إرسال تفاصيل الفاتورة إلى ahmad نقطة b آت gmail نقطة co نقطة id، المجموع النهائي هو أربعة وعشرون ألفاً وأربعمائة وواحد وتسعون دولاراً."}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: www.bank.my"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: w w w نقطة bank نقطة my"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: https://sltnet.lk"}, {"role": "assistant", "content": "يرجى تأكيد استلام الطرد عبر الضغط على الرابط التالي: sltnet نقطة lk"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي REF-٢٠٢٤-٠١١٣ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي REF اثنان صفر اثنان أربعة صفر واحد واحد ثلاثة للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو A١٢ وسنقوم بمعالجته خلال ٦ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو A واحد اثنان وسنقوم بمعالجته خلال ستة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق 747428 لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق سبعة أربعة سبعة أربعة اثنان ثمانية لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو INV0472 وسنقوم بمعالجته خلال 5 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو INV صفر أربعة سبعة اثنان وسنقوم بمعالجته خلال خمسة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو WXY 1234 وسنقوم بمعالجته خلال 113 أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو WXY واحد اثنان ثلاثة أربعة وسنقوم بمعالجته خلال مائة وثلاثة عشر أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال الرقم التعريفي A١٢ للدخول إلى بوابة الطالب"}, {"role": "assistant", "content": "يرجى إدخال الرقم التعريفي A واحد اثنان للدخول إلى بوابة الطالب"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "أهلاً بك، رقم طلبك هو B٢-١٤ وسنقوم بمعالجته خلال ٦ أيام عمل"}, {"role": "assistant", "content": "أهلاً بك، رقم طلبك هو B اثنان واحد أربعة وسنقوم بمعالجته خلال ستة أيام عمل"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إدخال رمز التحقق ٣٦٧٩٧ لإتمام عملية تسجيل الدخول"}, {"role": "assistant", "content": "يرجى إدخال رمز التحقق ثلاثة ستة سبعة تسعة سبعة لإتمام عملية تسجيل الدخول"}], "lang": "ar", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 25 - 28."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-cinq à vingt-huit."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 89% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-neuf pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 20 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de vingt mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 8 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de huit mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse kumar75@posta.it."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse kumar sept cinq arobase posta point it."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 34 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de trente-quatre mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 1-3."}, {"role": "assistant", "content": "L'agence sera ouverte demain de un à trois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 14e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil quatorzième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 26ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil vingt-sixième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 60,8 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de soixante virgule huit pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 5-14."}, {"role": "assistant", "content": "L'agence sera ouverte demain de cinq à quatorze."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 43 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de quarante-trois mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 84% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-quatre pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 25 - 27."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-cinq à vingt-sept."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 12-14."}, {"role": "assistant", "content": "L'agence sera ouverte demain de douze à quatorze."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 16 - 17."}, {"role": "assistant", "content": "L'agence sera ouverte demain de seize à dix-sept."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse ravi@gmail.co.id."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse ravi arobase gmail point co point id."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 41,6% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quarante et un virgule six pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 2e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil deuxième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 29–31."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-neuf à trente et un."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 4 - 7."}, {"role": "assistant", "content": "L'agence sera ouverte demain de quatre à sept."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 17ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil dix-septième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 2ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil deuxième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 5 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de cinq mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 91 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-onze pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia17@gmail.co.id."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia un sept arobase gmail point co point id."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 47% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quarante-sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 56% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante-six pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 30–38."}, {"role": "assistant", "content": "L'agence sera ouverte demain de trente à trente-huit."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 738 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de sept cent trente-huit mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse anna@outlook.sa."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse anna arobase outlook point sa."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 30e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil trentième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 1ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil premier étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 9e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil neuvième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 18e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil dix-huitième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 90 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-dix pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 28,7% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de vingt-huit virgule sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 7 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de sept mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 8ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil huitième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 91 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de quatre-vingt-onze mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse pedro@mail.fr."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse pedro arobase mail point fr."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 27 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de vingt-sept mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 18 - 22."}, {"role": "assistant", "content": "L'agence sera ouverte demain de dix-huit à vingt-deux."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse info45@sltnet.lk."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse info quatre cinq arobase sltnet point lk."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia4@yahoo.com."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia quatre arobase yahoo point com."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 23ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil vingt-troisième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 3% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de trois pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 11-20."}, {"role": "assistant", "content": "L'agence sera ouverte demain de onze à vingt."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 2 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de deux mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 17–21."}, {"role": "assistant", "content": "L'agence sera ouverte demain de dix-sept à vingt et un."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse nurul_h43@onet.pl."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse nurul underscore h quatre trois arobase onet point pl."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 722 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de sept cent vingt-deux mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 30 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de trente mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse kumar@posta.it."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse kumar arobase posta point it."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 297 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de deux cent quatre-vingt-dix-sept mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 20,4 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de vingt virgule quatre pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 7e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil septième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 153 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de cent cinquante-trois mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 61% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de soixante et un pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 4 - 11."}, {"role": "assistant", "content": "L'agence sera ouverte demain de quatre à onze."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse fatima@sltnet.lk."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse fatima arobase sltnet point lk."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 1er étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil premier étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 2 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de deux pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 50,7% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante virgule sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia@company.co."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse sofia arobase company point co."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 50% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse mei.ling63@outlook.sa."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse mei point ling six trois arobase outlook point sa."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 3 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de trois mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 21% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de vingt et un pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 2-3."}, {"role": "assistant", "content": "L'agence sera ouverte demain de deux à trois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 55% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante-cinq pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 7,7% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de sept virgule sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 16-21."}, {"role": "assistant", "content": "L'agence sera ouverte demain de seize à vingt et un."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 10ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil dixième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing@example.com."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing arobase example point com."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 95,5% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-quinze virgule cinq pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 66% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de soixante-six pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 19-21."}, {"role": "assistant", "content": "L'agence sera ouverte demain de dix-neuf à vingt et un."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 5ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil cinquième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 30-38."}, {"role": "assistant", "content": "L'agence sera ouverte demain de trente à trente-huit."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 41 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de quarante et un mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 23e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil vingt-troisième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse chen88@bank.my."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse chen huit huit arobase bank point my."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 3-11."}, {"role": "assistant", "content": "L'agence sera ouverte demain de trois à onze."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 30ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil trentième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 97% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-dix-sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 50,7 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante virgule sept pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 1 - 4."}, {"role": "assistant", "content": "L'agence sera ouverte demain de un à quatre."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse jan.devries81@sltnet.lk."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse jan point devries huit un arobase sltnet point lk."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing@sltnet.lk."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing arobase sltnet point lk."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 27–30."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-sept à trente."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 15-23."}, {"role": "assistant", "content": "L'agence sera ouverte demain de quinze à vingt-trois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 69 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de soixante-neuf mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse info@web.de."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse info arobase web point de."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 16 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de seize mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 84,3% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de quatre-vingt-quatre virgule trois pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 18–20."}, {"role": "assistant", "content": "L'agence sera ouverte demain de dix-huit à vingt."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 49 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de quarante-neuf mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 70% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de soixante-dix pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 15e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil quinzième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 51% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de cinquante et un pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 22,3% cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de vingt-deux virgule trois pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 28-29."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-huit à vingt-neuf."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse ali@company.co."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse ali arobase company point co."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 20e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil vingtième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse juan@bank.my."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse juan arobase bank point my."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 9ème étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil neuvième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse jean.dupont@gmail.co.id."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse jean point dupont arobase gmail point co point id."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 26 - 32."}, {"role": "assistant", "content": "L'agence sera ouverte demain de vingt-six à trente-deux."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'agence sera ouverte demain de 11-15."}, {"role": "assistant", "content": "L'agence sera ouverte demain de onze à quinze."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing@company.co."}, {"role": "assistant", "content": "Nous avons bien reçu votre demande envoyée depuis l'adresse billing arobase company point co."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux de réussite à l'examen est de 61 % cette année."}, {"role": "assistant", "content": "Le taux de réussite à l'examen est de soixante et un pour cent cette année."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 641 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de six cent quarante et un mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de nous fournir une pièce d'identité datant de moins de 312 mois."}, {"role": "assistant", "content": "Merci de nous fournir une pièce d'identité datant de moins de trois cent douze mois."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de vous présenter à l'accueil 26e étage du bâtiment B."}, {"role": "assistant", "content": "Merci de vous présenter à l'accueil vingt-sixième étage du bâtiment B."}], "lang": "fr", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 14/10/2016 en el horario de 23 - 27."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el catorce de octubre de dos mil dieciséis en el horario de veintitrés a veintisiete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 8 - 11 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de ocho a once días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 49 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cuarenta y nueve dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 23 - 30 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintitrés a treinta días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo A12 a las 4.15."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo A uno dos a las cuatro quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 2 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los dos dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo PO 55219 a las 14:10."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo PO cinco cinco dos uno nueve a las catorce diez."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 11 - 13 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de once a trece días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 25/05/2015 en el horario de 1–4."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veinticinco de mayo de dos mil quince en el horario de uno a cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo INV0472 a las 23:10."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo INV cero cuatro siete dos a las veintitrés diez."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo A12 a las 19:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo A uno dos a las diecinueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 7-10 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de siete a diez días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 1.º día del mes de 1983."}, {"role": "assistant", "content": "Su cita ha sido programada para el primero día del mes de mil novecientos ochenta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC1234 a las 20:15."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC uno dos tres cuatro a las veinte quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 21/12/2030 en el horario de 11–18."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintiuno de diciembre de dos mil treinta en el horario de once a dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 29 - 38 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintinueve a treinta y ocho días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 17º día del mes de 1950."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo séptimo día del mes de mil novecientos cincuenta."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 15 de septiembre en el horario de 7-15."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el quince de septiembre en el horario de siete a quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 29-38 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintinueve a treinta y ocho días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 4 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cuatro dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 9 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los nueve dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 16/9/2004 en el horario de 29 - 35."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el dieciséis de septiembre de dos mil cuatro en el horario de veintinueve a treinta y cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo PO 55219 a las 3:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo PO cinco cinco dos uno nueve a las tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 6/5/2000 en el horario de 29-36."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el seis de mayo de dos mil en el horario de veintinueve a treinta y seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 25 - 31 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veinticinco a treinta y uno días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 28-30 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintiocho a treinta días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 14-24 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de catorce a veinticuatro días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 23 de julio en el horario de 17–18."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintitrés de julio en el horario de diecisiete a dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 20/1/2030 en el horario de 21 - 24."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veinte de enero de dos mil treinta en el horario de veintiuno a veinticuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 351 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los trescientos cincuenta y uno dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo B2-14 a las 23.15."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo B dos uno cuatro a las veintitrés quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 5 de diciembre de 1998 en el horario de 9 - 11."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el cinco de diciembre de mil novecientos noventa y ocho en el horario de nueve a once."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 30º día del mes de 1960."}, {"role": "assistant", "content": "Su cita ha sido programada para el trigésimo día del mes de mil novecientos sesenta."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 10/09/2000 en el horario de 25 - 27."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el diez de septiembre de dos mil en el horario de veinticinco a veintisiete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 18 - 28 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de dieciocho a veintiocho días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 14.º día del mes de 2018."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo cuarto día del mes de dos mil dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 13º día del mes de 1991."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo tercero día del mes de mil novecientos noventa y uno."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ8501 a las 10:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ ocho cinco cero uno a las diez."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ8501 a las 6:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ ocho cinco cero uno a las seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo X-12340567 a las 10.45."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo X uno dos tres cuatro cero cinco seis siete a las diez cuarenta y cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 21-27 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintiuno a veintisiete días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 9 - 17 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de nueve a diecisiete días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 4 de diciembre en el horario de 9-15."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el cuatro de diciembre en el horario de nueve a quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 22 de septiembre en el horario de 20-27."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintidós de septiembre en el horario de veinte a veintisiete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 22–27 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintidós a veintisiete días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 306 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los trescientos seis dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 9–13 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de nueve a trece días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 697 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los seiscientos noventa y siete dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 22º día del mes de 1959."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo segundo día del mes de mil novecientos cincuenta y nueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 25º día del mes de 1997."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo quinto día del mes de mil novecientos noventa y siete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC1234 a las 18:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC uno dos tres cuatro a las dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC1234 a las 4:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC uno dos tres cuatro a las cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 12 de mayo de 2002 en el horario de 20 - 23."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el doce de mayo de dos mil dos en el horario de veinte a veintitrés."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 4º día del mes de 2004."}, {"role": "assistant", "content": "Su cita ha sido programada para el cuarto día del mes de dos mil cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 23.º día del mes de 1983."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo tercero día del mes de mil novecientos ochenta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 21 - 29 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintiuno a veintinueve días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo INV0472 a las 14.40."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo INV cero cuatro siete dos a las catorce cuarenta."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 24-28 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veinticuatro a veintiocho días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 21º día del mes de 2031."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo primero día del mes de dos mil treinta y uno."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 5-13 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de cinco a trece días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 13º día del mes de 1956."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo tercero día del mes de mil novecientos cincuenta y seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC1234 a las 8.30."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC uno dos tres cuatro a las ocho treinta."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 12/5/2014 en el horario de 15-18."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el doce de mayo de dos mil catorce en el horario de quince a dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo REF-2024-0113 a las 11:05."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo REF dos cero dos cuatro cero uno uno tres a las once cero cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 543 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los quinientos cuarenta y tres dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 29.º día del mes de 2033."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo noveno día del mes de dos mil treinta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 15 - 20 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de quince a veinte días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 14.º día del mes de 1957."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo cuarto día del mes de mil novecientos cincuenta y siete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY 1234 a las 4:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY uno dos tres cuatro a las cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 6–13 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de seis a trece días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 62 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los sesenta y dos dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 19 de septiembre en el horario de 24-31."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el diecinueve de septiembre en el horario de veinticuatro a treinta y uno."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 10 de junio en el horario de 27-29."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el diez de junio en el horario de veintisiete a veintinueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 23.º día del mes de 1957."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo tercero día del mes de mil novecientos cincuenta y siete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 29º día del mes de 1966."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo noveno día del mes de mil novecientos sesenta y seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 3 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los tres dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 23–33 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintitrés a treinta y tres días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 31.º día del mes de 1956."}, {"role": "assistant", "content": "Su cita ha sido programada para el trigésimo primero día del mes de mil novecientos cincuenta y seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 24 de noviembre en el horario de 19-22."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veinticuatro de noviembre en el horario de diecinueve a veintidós."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 8 - 12 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de ocho a doce días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 14.º día del mes de 1953."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo cuarto día del mes de mil novecientos cincuenta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY 1234 a las 2:15."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY uno dos tres cuatro a las dos quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 9º día del mes de 1973."}, {"role": "assistant", "content": "Su cita ha sido programada para el noveno día del mes de mil novecientos setenta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 816 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los ochocientos dieciséis dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 8 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los ocho dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 24º día del mes de 1984."}, {"role": "assistant", "content": "Su cita ha sido programada para el vigésimo cuarto día del mes de mil novecientos ochenta y cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 10-16 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de diez a dieciséis días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo B2-14 a las 17:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo B dos uno cuatro a las diecisiete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 16 de abril de 2004 en el horario de 26–36."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el dieciséis de abril de dos mil cuatro en el horario de veintiséis a treinta y seis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo MH370 a las 12:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo MH tres siete cero a las doce."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 15º día del mes de 1983."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo quinto día del mes de mil novecientos ochenta y tres."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 30-40 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de treinta a cuarenta días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 3 de febrero en el horario de 9-14."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el tres de febrero en el horario de nueve a catorce."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC1234 a las 19:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo ABC uno dos tres cuatro a las diecinueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 961 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los novecientos sesenta y uno dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ8501 a las 2:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ ocho cinco cero uno a las dos."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 20 de noviembre en el horario de 26–28."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veinte de noviembre en el horario de veintiséis a veintiocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ8501 a las 9.00."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo QZ ocho cinco cero uno a las nueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 18 de diciembre de 2009 en el horario de 21-28."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el dieciocho de diciembre de dos mil nueve en el horario de veintiuno a veintiocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 478 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cuatrocientos setenta y ocho dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 26 de mayo de 1995 en el horario de 6-16."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintiséis de mayo de mil novecientos noventa y cinco en el horario de seis a dieciséis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 08/04/2001 en el horario de 2-5."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el ocho de abril de dos mil uno en el horario de dos a cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 53 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cincuenta y tres dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo TK-88213 a las 4.00."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo TK ocho ocho dos uno tres a las cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo REF-2024-0113 a las 19.45."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo REF dos cero dos cuatro cero uno uno tres a las diecinueve cuarenta y cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 23/10/2029 en el horario de 19 - 24."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintitrés de octubre de dos mil veintinueve en el horario de diecinueve a veinticuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 47 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cuarenta y siete dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo AK6120 a las 9.45."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo AK seis uno dos cero a las nueve cuarenta y cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 12-17 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de doce a diecisiete días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 5 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los cinco dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 12º día del mes de 1979."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimosegundo día del mes de mil novecientos setenta y nueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 28/4/1999 en el horario de 12–22."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veintiocho de abril de mil novecientos noventa y nueve en el horario de doce a veintidós."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 15º día del mes de 1994."}, {"role": "assistant", "content": "Su cita ha sido programada para el décimo quinto día del mes de mil novecientos noventa y cuatro."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 25-30 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veinticinco a treinta días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 10/05/2010 en el horario de 17 - 18."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el diez de mayo de dos mil diez en el horario de diecisiete a dieciocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 25/12/1999 en el horario de 10 - 16."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el veinticinco de diciembre de mil novecientos noventa y nueve en el horario de diez a dieciséis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 08/01/2000 en el horario de 25-29."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el ocho de enero de dos mil en el horario de veinticinco a veintinueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 4 de julio de 2007 en el horario de 25 - 28."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el cuatro de julio de dos mil siete en el horario de veinticinco a veintiocho."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 6 de septiembre en el horario de 6-7."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el seis de septiembre en el horario de seis a siete."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY 1234 a las 18.15."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo WXY uno dos tres cuatro a las dieciocho quince."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo A12 a las 13:00 h."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo A uno dos a las trece."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo AK6120 a las 2.05."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo AK seis uno dos cero a las dos cero cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 27–36 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de veintisiete a treinta y seis días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hemos detectado un inicio de sesión desde el dispositivo X-12340567 a las 1:10."}, {"role": "assistant", "content": "Hemos detectado un inicio de sesión desde el dispositivo X uno dos tres cuatro cero cinco seis siete a las uno diez."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 6 de septiembre de 2009 en el horario de 17 - 22."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el seis de septiembre de dos mil nueve en el horario de diecisiete a veintidós."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El plazo de entrega estimado es de 3–6 días hábiles."}, {"role": "assistant", "content": "El plazo de entrega estimado es de tres a seis días hábiles."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para completar el registro, ingrese los 344 dígitos de su identificación."}, {"role": "assistant", "content": "Para completar el registro, ingrese los trescientos cuarenta y cuatro dígitos de su identificación."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 10/12/2022 en el horario de 24 - 26."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el diez de diciembre de dos mil veintidós en el horario de veinticuatro a veintiséis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 1º día del mes de 2029."}, {"role": "assistant", "content": "Su cita ha sido programada para el primero día del mes de dos mil veintinueve."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Su cita ha sido programada para el 5º día del mes de 1955."}, {"role": "assistant", "content": "Su cita ha sido programada para el quinto día del mes de mil novecientos cincuenta y cinco."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El técnico visitará tu domicilio el 12/01/2027 en el horario de 12-16."}, {"role": "assistant", "content": "El técnico visitará tu domicilio el doce de enero de dos mil veintisiete en el horario de doce a dieciséis."}], "lang": "es", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 16. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim sechzehnte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 92,59€ bis zum 25. Februar 2014."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von zweiundneunzig Euro und neunundfünfzig Cent bis zum fünfundzwanzigsten Februar zweitausendvierzehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1m Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Meter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 3. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim dritte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 26. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim sechsundzwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 177 88948693 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben sieben acht acht neun vier acht sechs neun drei an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 172 94363838 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben zwei neun vier drei sechs drei acht drei acht an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 175 01972545 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben fünf null eins neun sieben zwei fünf vier fünf an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 39 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa neununddreißig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 269,90€ bis zum 13.07.2000."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von zweihundertneunundsechzig Euro und neunzig Cent bis zum dreizehnten Juli zweitausend."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 81 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa einundachtzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0177 53768831 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins sieben sieben fünf drei sieben sechs acht acht drei eins an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 29. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim neunundzwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 177 45048449 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben sieben vier fünf null vier acht vier vier neun an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 6. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim sechste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1cm Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Zentimeter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 7 m Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits sieben Meter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 6 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa sechs Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0179 80521776 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins sieben neun acht null fünf zwei eins sieben sieben sechs an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 57 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa siebenundfünfzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 119,90€ bis zum 25.11.2027."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von einhundertneunzehn Euro und neunzig Cent bis zum fünfundzwanzigsten November zweitausendsiebenundzwanzig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0153 23596728 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf drei zwei drei fünf neun sechs sieben zwei acht an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 8,1cm Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits acht Komma eins Zentimeter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 8,5m Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits acht Komma fünf Meter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0174 17449065 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins sieben vier eins sieben vier vier neun null sechs fünf an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0150 30755306 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf null drei null sieben fünf fünf drei null sechs an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0150 24211850 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf null zwei vier zwei eins eins acht fünf null an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0152 98614837 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf zwei neun acht sechs eins vier acht drei sieben an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 10. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim zehnte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 398kg Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits dreihundertachtundneunzig Kilogramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0157 47270735 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf sieben vier sieben zwei sieben null sieben drei fünf an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 195,2g Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits einhundertfünfundneunzig Komma zwei Gramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 2. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim zweite Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 644 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa sechshundertvierundvierzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 88,90 € bis zum 19. Dezember 2030."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von achtundachtzig Euro und neunzig Cent bis zum neunzehnten Dezember zweitausenddreißig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 29 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa neunundzwanzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 71.431,50€ bis zum 21.2.2015."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von einundsiebzigtausendvierhunderteinunddreißig Euro und fünfzig Cent bis zum einundzwanzigsten Februar zweitausendfünfzehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 176 56556122 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben sechs fünf sechs fünf fünf sechs eins zwei zwei an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 669 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa sechshundertneunundsechzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 174 22278350 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben vier zwei zwei zwei sieben acht drei fünf null an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 89.057€ bis zum 27. August."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von neunundachtzigtausendsiebenundfünfzig Euro bis zum siebenundzwanzigsten August."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 030 4736644 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null drei null vier sieben drei sechs sechs vier vier an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 9 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa neun Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 210 € bis zum 1. März."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von zweihundertzehn Euro bis zum ersten März."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 14 € bis zum 22.04.2009."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von vierzehn Euro bis zum zweiundzwanzigsten April zweitausendneun."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 89.909,76 bis zum 11. Oktober."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von neunundachtzigtausendneunhundertneun Euro und sechsundsiebzig Cent bis zum elften Oktober."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 3 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa drei Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 4. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim vierte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 29.268,50 bis zum 2. Februar 2018."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von neunundzwanzigtausendzweihundertachtundsechzig Euro und fünfzig Cent bis zum zweiten Februar zweitausendachtzehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 030 9586306 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null drei null neun fünf acht sechs drei null sechs an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 18,1km Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits achtzehn Komma eins Kilometer Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0156 87451336 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf sechs acht sieben vier fünf eins drei drei sechs an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 63.387,90 € bis zum 21.05.2017."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von dreiundsechzigtausenddreihundertsiebenundachtzig Euro und neunzig Cent bis zum einundzwanzigsten Mai zweitausendsiebzehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0173 21068713 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins sieben drei zwei eins null sechs acht sieben eins drei an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 030 5881491 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null drei null fünf acht acht eins vier neun eins an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 13. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim dreizehnte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1 kg Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Kilogramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 640,90 bis zum 02.05.2003."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von sechshundertvierzig Euro und neunzig Cent bis zum zweiten Mai zweitausenddrei."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1 ml Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Milliliter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 19,49 bis zum 7. Dezember."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von neunzehn Euro und neunundvierzig Cent bis zum siebten Dezember."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1,4 °C Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits eins Komma vier Grad Celsius Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1 °C Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Grad Celsius Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 464,50€ bis zum 27. Januar."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von vierhundertvierundsechzig Euro und fünfzig Cent bis zum siebenundzwanzigsten Januar."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 5. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim fünfte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1,6g Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits eins Komma sechs Gramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 72.145,50 € bis zum 03.03.2020."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von zweiundsiebzigtausendeinhundertfünfundvierzig Euro und fünfzig Cent bis zum dritten März zweitausendzwanzig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 46€ bis zum 02.06.2030."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von sechsundvierzig Euro bis zum zweiten Juni zweitausenddreißig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 32.168,50 € bis zum 22. März 2021."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von zweiunddreißigtausendeinhundertachtundsechzig Euro und fünfzig Cent bis zum zweiundzwanzigsten März zweitausendeinundzwanzig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 352 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa dreihundertzweiundfünfzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 28. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim achtundzwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1l Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Liter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 145,90 bis zum 14. Juli."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von einhundertfünfundvierzig Euro und neunzig Cent bis zum vierzehnten Juli."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 17. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim siebzehnte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 177 56156987 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben sieben fünf sechs eins fünf sechs neun acht sieben an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 27. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim siebenundzwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 9 ml Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits neun Milliliter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 15. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim fünfzehnte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 030 0741565 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null drei null null sieben vier eins fünf sechs fünf an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 172 70477305 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben zwei sieben null vier sieben sieben drei null fünf an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 59 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa neunundfünfzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 178 63724088 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben acht sechs drei sieben zwei vier null acht acht an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 451 bis zum 1.1.2020."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von vierhunderteinundfünfzig Euro bis zum ersten Januar zweitausendzwanzig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 178 48817116 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben acht vier acht acht eins sieben eins eins sechs an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 742 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa siebenhundertzweiundvierzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 12. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim zwölfte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 71 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa einundsiebzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 11. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim elfte Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 31 bis zum 4. Dezember 2017."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von einunddreißig Euro bis zum vierten Dezember zweitausendsiebzehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 31m Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits einunddreißig Meter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0178 13873152 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins sieben acht eins drei acht sieben drei eins fünf zwei an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 7 ml Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits sieben Milliliter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 267 ml Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits zweihundertsiebenundsechzig Milliliter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 356,4 cm Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits dreihundertsechsundfünfzig Komma vier Zentimeter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter 0154 79543253 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter null eins fünf vier sieben neun fünf vier drei zwei fünf drei an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 20. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim zwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1 cm Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Zentimeter Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 608,50€ bis zum 11. September 2011."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von sechshundertacht Euro und fünfzig Cent bis zum elften September zweitausendelf."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 453 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa vierhundertdreiundfünfzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 561 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa fünfhunderteinundsechzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 93,50 bis zum 21. August 2013."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von dreiundneunzig Euro und fünfzig Cent bis zum einundzwanzigsten August zweitausenddreizehn."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 55 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa fünfundfünfzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von EUR 60,90 bis zum 24. Januar."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von sechzig Euro und neunzig Cent bis zum vierundzwanzigsten Januar."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 8kg Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits acht Kilogramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 2 g Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits zwei Gramm Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 364 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa dreihundertvierundsechzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 18 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa achtzehn Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte rufen Sie uns unter +49 177 75122659 an, falls Sie Ihren Termin verschieben müssen."}, {"role": "assistant", "content": "Bitte rufen Sie uns unter plus vier neun eins sieben sieben sieben fünf eins zwei zwei sechs fünf neun an, falls Sie Ihren Termin verschieben müssen."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 719 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa siebenhundertneunzehn Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte überweisen Sie den Betrag von 611,10 € bis zum 19.2.2023."}, {"role": "assistant", "content": "Bitte überweisen Sie den Betrag von sechshundertelf Euro und zehn Cent bis zum neunzehnten Februar zweitausenddreiundzwanzig."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Willkommen beim 22. Monat Ihrer kostenlosen Testphase."}, {"role": "assistant", "content": "Willkommen beim zweiundzwanzigste Monat Ihrer kostenlosen Testphase."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 45 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa fünfundvierzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben festgestellt, dass Sie bereits 1 km Ihres Volumens verbraucht haben."}, {"role": "assistant", "content": "Wir haben festgestellt, dass Sie bereits ein Kilometer Ihres Volumens verbraucht haben."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Wartezeit beträgt derzeit etwa 49 Minuten."}, {"role": "assistant", "content": "Die Wartezeit beträgt derzeit etwa neunundvierzig Minuten."}], "lang": "de", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.company.co/login."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto company punto co barra login."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 4%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quattro per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 22,24 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ventidue virgola due quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 91%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del novantuno per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.sapo.pt/login."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto sapo punto pt barra login."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 888 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai ottocentottantotto messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 977 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno novecentosettantasette ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://www.yahoo.com/support."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto yahoo punto com barra support."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 33 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno trentatré ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 20%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del venti per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 5 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno cinque ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 440 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai quattrocentoquaranta messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 18%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del diciotto per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 774 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai settecentosettantaquattro messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 6 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai sei messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 49%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quarantanove per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 936 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno novecentotrentasei ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 8,4 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del otto virgola quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 197 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno centonovantasette ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 26,26 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ventisei virgola due sei percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 45,8%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quarantacinque virgola otto per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://www.example.com/support."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto example punto com barra support."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://web.de."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo web punto de."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 4 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai quattro messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.sltnet.lk/login."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto sltnet punto lk barra login."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 952 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno novecentocinquantadue ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.company.co."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto company punto co."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 97%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del novantasette per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 62 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno sessantadue ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 8,9%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del otto virgola nove per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 20 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai venti messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 6 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno sei ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 26 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai ventisei messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 8%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del otto per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 74,3%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del settantaquattro virgola tre per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 67%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sessantasette per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 16,39 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del sedici virgola tre nove percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 7,24 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del sette virgola due quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://www.company.co/support."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto company punto co barra support."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 99,55 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del novantanove virgola cinque cinque percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 79,35 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del settantanove virgola tre cinque percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 67,59 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del sessantasette virgola cinque nove percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 16,1 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del sedici virgola uno percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.sltnet.lk."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto sltnet punto lk."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 9 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai nove messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 83%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del ottantatré per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 8 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai otto messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 66 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno sessantasei ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 65,2%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sessantacinque virgola due per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 48,3 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del quarantotto virgola tre percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 79 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai settantanove messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 77,4%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del settantasette virgola quattro per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 6%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sei per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 65%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sessantacinque per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 53%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del cinquantatré per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 10,3 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del dieci virgola tre percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 324 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai trecentoventiquattro messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 2 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno due ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 160 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai centosessanta messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 48,2%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quarantotto virgola due per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://bank.my/help."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo bank punto my barra help."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://company.co."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo company punto co."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 70%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del settanta per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 68%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sessantotto per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 3,74 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del tre virgola sette quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 8 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno otto ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 38,56 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del trentotto virgola cinque sei percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 77,42 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del settantasette virgola quattro due percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 87,21 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ottantasette virgola due uno percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 66%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del sessantasei per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 17%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del diciassette per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 81,4 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ottantuno virgola quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://scicom.com.my/faq."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo scicom punto com punto my barra faq."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 45,53 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del quarantacinque virgola cinque tre percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 3,9 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del tre virgola nove percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 84,19 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ottantaquattro virgola uno nove percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 14 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai quattordici messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 371 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai trecentosettantuno messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 511 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno cinquecentoundici ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 10,4 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del dieci virgola quattro percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 79,28 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del settantanove virgola due otto percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 99,4%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del novantanove virgola quattro per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 46,18 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del quarantasei virgola uno otto percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 9 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno nove ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo www.outlook.sa."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo w w w punto outlook punto sa."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 3,7 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del tre virgola sette percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 45%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quarantacinque per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://web.de/help."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo web punto de barra help."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 43%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del quarantatré per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 653 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai seicentocinquantatre messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 12%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del dodici per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 52 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno cinquantadue ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per scaricare il materiale didattico, visita l'indirizzo https://gmail.com/help."}, {"role": "assistant", "content": "Per scaricare il materiale didattico, visita l'indirizzo gmail punto com barra help."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 46,30 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del quarantasei virgola tre zero percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 26,40 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ventisei virgola quattro zero percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 93,8 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del novantatré virgola otto percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 54 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno cinquantaquattro ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 313 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno trecentotredici ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 50,6 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del cinquanta virgola sei percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 465 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno quattrocentosessantacinque ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 57 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno cinquantasette ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 28 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai ventotto messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le ricordiamo di rimanere a digiuno per almeno 3 ore prima del prelievo."}, {"role": "assistant", "content": "Le ricordiamo di rimanere a digiuno per almeno tre ore prima del prelievo."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai superato la soglia di traffico consentita del 36,4%."}, {"role": "assistant", "content": "Hai superato la soglia di traffico consentita del trentasei virgola quattro per cento."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 55,51 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del cinquantacinque virgola cinque uno percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 13,63 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del tredici virgola sei tre percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 589 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai cinquecentottantanove messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 2 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai due messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hai 51 messaggi in attesa di lettura."}, {"role": "assistant", "content": "Hai cinquantuno messaggi in attesa di lettura."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo investimento è cresciuto del 23,72 percento nell'ultimo mese."}, {"role": "assistant", "content": "Il tuo investimento è cresciuto del ventitré virgola sette due percento nell'ultimo mese."}], "lang": "it", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 9.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O nono pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 925228 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação nove dois cinco dois dois oito para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 12 de abril e o valor total é R$ 59."}, {"role": "assistant", "content": "Sua fatura vence no dia doze de abril e o valor total é cinquenta e nove reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 80182 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação oito zero um oito dois para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 91796 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação nove um sete nove seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 259848 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é dois cinco nove oito quatro oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 22:20 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às vinte e dois horas e vinte de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 836417 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é oito três seis quatro um sete para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 16469 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação um seis quatro seis nove para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 9687 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é nove seis oito sete para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 1/10/2011 e o valor total é R$ 56.522,90."}, {"role": "assistant", "content": "Sua fatura vence no dia um de outubro de dois mil e onze e o valor total é cinquenta e seis mil quinhentos e vinte e dois reais e noventa centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 30631 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação três zero seis três um para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 24/10/2002 e o valor total é R$1,90."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e quatro de outubro de dois mil e dois e o valor total é um real e noventa centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 30941 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é três zero nove quatro um para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 1º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O primeiro pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 783348 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação sete oito três três quatro oito para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 0:00 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às zero horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 9 de dezembro de 2003 e o valor total é 11.051,50 €."}, {"role": "assistant", "content": "Sua fatura vence no dia nove de dezembro de dois mil e três e o valor total é onze mil e cinquenta e um euros e cinquenta cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 10h30 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dez horas e trinta de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 30.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O trigésimo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 4354 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação quatro três cinco quatro para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 03/12/2022 e o valor total é EUR 365."}, {"role": "assistant", "content": "Sua fatura vence no dia três de dezembro de dois mil e vinte e dois e o valor total é trezentos e sessenta e cinco euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 13.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo terceiro pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 3/5/2018 e o valor total é 758,90€."}, {"role": "assistant", "content": "Sua fatura vence no dia três de maio de dois mil e dezoito e o valor total é setecentos e cinquenta e oito euros e noventa cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 7:00 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às sete horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 641924 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação seis quatro um nove dois quatro para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 7h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às sete horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 4209 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é quatro dois zero nove para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 15º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo quinto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 258780 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação dois cinco oito sete oito zero para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 5h45 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às cinco horas e quarenta e cinco de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 5839 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação cinco oito três nove para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 19.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo nono pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 561348 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação cinco seis um três quatro oito para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 428822 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação quatro dois oito oito dois dois para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 056108 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é zero cinco seis um zero oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 14 de setembro e o valor total é R$51."}, {"role": "assistant", "content": "Sua fatura vence no dia catorze de setembro e o valor total é cinquenta e um reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 28096 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é dois oito zero nove seis para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 28345 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação dois oito três quatro cinco para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 22h00 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às vinte e dois horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 10º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 12º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo segundo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 17358 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação um sete três cinco oito para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 0005 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é zero zero zero cinco para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 28/8/2003 e o valor total é R$97.337."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e oito de agosto de dois mil e três e o valor total é noventa e sete mil trezentos e trinta e sete reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 11 de março de 2010 e o valor total é EUR 552,50."}, {"role": "assistant", "content": "Sua fatura vence no dia onze de março de dois mil e dez e o valor total é quinhentos e cinquenta e dois euros e cinquenta cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 19 de dezembro e o valor total é 10.335€."}, {"role": "assistant", "content": "Sua fatura vence no dia dezanove de dezembro e o valor total é dez mil trezentos e trinta e cinco euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 11h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às onze horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 19h15 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dezanove horas e quinze de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 20º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 23 de junho e o valor total é 93.700€."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e três de junho e o valor total é noventa e três mil e setecentos euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 02352 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação zero dois três cinco dois para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 6/7/2013 e o valor total é R$20.733."}, {"role": "assistant", "content": "Sua fatura vence no dia seis de julho de dois mil e treze e o valor total é vinte mil setecentos e trinta e três reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 04/07/2015 e o valor total é 98€."}, {"role": "assistant", "content": "Sua fatura vence no dia quatro de julho de dois mil e quinze e o valor total é noventa e oito euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 25 de junho e o valor total é R$ 411."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e cinco de junho e o valor total é quatrocentos e onze reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 22/06/2023 e o valor total é 39€."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e dois de junho de dois mil e vinte e três e o valor total é trinta e nove euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 13/02/2012 e o valor total é EUR 18."}, {"role": "assistant", "content": "Sua fatura vence no dia treze de fevereiro de dois mil e doze e o valor total é dezoito euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 22º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo segundo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 26º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo sexto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 1h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às um hora de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 12h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às doze horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 21h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às vinte e um horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 16.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo sexto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 3 de setembro e o valor total é 81.184,48 €."}, {"role": "assistant", "content": "Sua fatura vence no dia três de setembro e o valor total é oitenta e um mil cento e oitenta e quatro euros e quarenta e oito cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 4586 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação quatro cinco oito seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 03/06/2016 e o valor total é 704€."}, {"role": "assistant", "content": "Sua fatura vence no dia três de junho de dois mil e dezasseis e o valor total é setecentos e quatro euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 15h15 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às quinze horas e quinze de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 25 de maio e o valor total é R$97."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e cinco de maio e o valor total é noventa e sete reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 19 de janeiro e o valor total é 27,50€."}, {"role": "assistant", "content": "Sua fatura vence no dia dezanove de janeiro e o valor total é vinte e sete euros e cinquenta cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 403608 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é quatro zero três seis zero oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 18/8/2012 e o valor total é R$ 33,90."}, {"role": "assistant", "content": "Sua fatura vence no dia dezoito de agosto de dois mil e doze e o valor total é trinta e três reais e noventa centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 27 de outubro de 2016 e o valor total é 3.781,44€."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e sete de outubro de dois mil e dezasseis e o valor total é três mil setecentos e oitenta e um euros e quarenta e quatro cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 31406 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação três um quatro zero seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 19 de abril de 2010 e o valor total é 65.752€."}, {"role": "assistant", "content": "Sua fatura vence no dia dezanove de abril de dois mil e dez e o valor total é sessenta e cinco mil setecentos e cinquenta e dois euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 88676 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação oito oito seis sete seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 38686 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação três oito seis oito seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 9h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às nove horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 09/02/2028 e o valor total é R$ 78."}, {"role": "assistant", "content": "Sua fatura vence no dia nove de fevereiro de dois mil e vinte e oito e o valor total é setenta e oito reais."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 9h15 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às nove horas e quinze de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 301227 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação três zero um dois dois sete para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 5h05 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às cinco horas e cinco de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 23º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo terceiro pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 2183 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é dois um oito três para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 383925 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é três oito três nove dois cinco para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 2922 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação dois nove dois dois para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 24º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo quarto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 3868 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é três oito seis oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 42046 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação quatro dois zero quatro seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 2566 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação dois cinco seis seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 18 de novembro de 2023 e o valor total é EUR 20.031,72."}, {"role": "assistant", "content": "Sua fatura vence no dia dezoito de novembro de dois mil e vinte e três e o valor total é vinte mil e trinta e um euros e setenta e dois cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 16 de maio e o valor total é 863€."}, {"role": "assistant", "content": "Sua fatura vence no dia dezasseis de maio e o valor total é oitocentos e sessenta e três euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 15/07/2022 e o valor total é R$ 93,81."}, {"role": "assistant", "content": "Sua fatura vence no dia quinze de julho de dois mil e vinte e dois e o valor total é noventa e três reais e oitenta e um centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 18h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dezoito horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 700353 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é sete zero zero três cinco três para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 28º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo oitavo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 076110 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação zero sete seis um um zero para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 2:15 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dois horas e quinze de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 035624 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação zero três cinco seis dois quatro para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 20h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às vinte horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 6.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O sexto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 6:40 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às seis horas e quarenta de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 26/9/2026 e o valor total é 88,99 €."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e seis de setembro de dois mil e vinte e seis e o valor total é oitenta e oito euros e noventa e nove cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 655558 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é seis cinco cinco cinco cinco oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 115964 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é um um cinco nove seis quatro para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 16 de setembro de 1995 e o valor total é 39,90 €."}, {"role": "assistant", "content": "Sua fatura vence no dia dezasseis de setembro de mil novecentos e noventa e cinco e o valor total é trinta e nove euros e noventa cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 12.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo segundo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 569756 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação cinco seis nove sete cinco seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 15 de outubro e o valor total é R$34,59."}, {"role": "assistant", "content": "Sua fatura vence no dia quinze de outubro e o valor total é trinta e quatro reais e cinquenta e nove centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 16h20 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dezasseis horas e vinte de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 6º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O sexto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 2475 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é dois quatro sete cinco para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 482944 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é quatro oito dois nove quatro quatro para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 8786 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é oito sete oito seis para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 629430 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação seis dois nove quatro três zero para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 725892 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é sete dois cinco oito nove dois para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 6:45 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às seis horas e quarenta e cinco de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 18 de junho e o valor total é 49.437,90€."}, {"role": "assistant", "content": "Sua fatura vence no dia dezoito de junho e o valor total é quarenta e nove mil quatrocentos e trinta e sete euros e noventa cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 028176 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação zero dois oito um sete seis para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 74863 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação sete quatro oito seis três para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 26820 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é dois seis oito dois zero para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 91453 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação nove um quatro cinco três para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 04307 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação zero quatro três zero sete para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 24 de abril e o valor total é 833,50 €."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e quatro de abril e o valor total é oitocentos e trinta e três euros e cinquenta cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 90960 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação nove zero nove seis zero para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 4/10/2010 e o valor total é 96.670,90€."}, {"role": "assistant", "content": "Sua fatura vence no dia quatro de outubro de dois mil e dez e o valor total é noventa e seis mil seiscentos e setenta euros e noventa cêntimos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 961346 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é nove seis um três quatro seis para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 798695 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é sete nove oito seis nove cinco para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 1h00 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às um hora de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 7º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O sétimo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 6 de maio e o valor total é 75.484 €."}, {"role": "assistant", "content": "Sua fatura vence no dia seis de maio e o valor total é setenta e cinco mil quatrocentos e oitenta e quatro euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 882240 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação oito oito dois dois quatro zero para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 56392 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação cinco seis três nove dois para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 6h30 de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às seis horas e trinta de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 4h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às quatro horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 4118 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é quatro um um oito para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 18 de novembro e o valor total é R$ 81,51."}, {"role": "assistant", "content": "Sua fatura vence no dia dezoito de novembro e o valor total é oitenta e um reais e cinquenta e um centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 2h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dois horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 17h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dezassete horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 5.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O quinto pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, confirme se o seu CEP é 504326 para validarmos o endereço."}, {"role": "assistant", "content": "Por favor, confirme se o seu CEP é cinco zero quatro três dois seis para validarmos o endereço."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O motorista chegará ao seu endereço às 16h de amanhã."}, {"role": "assistant", "content": "O motorista chegará ao seu endereço às dezasseis horas de amanhã."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 43425 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação quatro três quatro dois cinco para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 21/04/2015 e o valor total é R$ 22.920,64."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e um de abril de dois mil e quinze e o valor total é vinte e dois mil novecentos e vinte reais e sessenta e quatro centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 20/9/1996 e o valor total é R$ 89,92."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte de setembro de mil novecentos e noventa e seis e o valor total é oitenta e nove reais e noventa e dois centavos."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 21º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O vigésimo primeiro pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O 17.º pagamento da sua parcela foi processado com sucesso."}, {"role": "assistant", "content": "O décimo sétimo pagamento da sua parcela foi processado com sucesso."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, informe o código de verificação 94017 para acessar seu prontuário."}, {"role": "assistant", "content": "Por favor, informe o código de verificação nove quatro zero um sete para acessar seu prontuário."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sua fatura vence no dia 27 de maio e o valor total é EUR 37."}, {"role": "assistant", "content": "Sua fatura vence no dia vinte e sete de maio e o valor total é trinta e sete euros."}], "lang": "pt", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 94 L."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is vierennegentig liter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 3-10 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen drie tot tien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 74,4% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar vierenzeventig komma vier procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 311,8cm."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is driehonderdelf komma acht centimeter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 153,3 km."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is honderddrieënvijftig komma drie kilometer."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 658.825 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zeshonderdachtenvijftigduizendachthonderdvijfentwintig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 8–9 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen acht tot negen bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 1°C."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is één graad Celsius."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1955 met 48%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vijfenvijftig met achtenveertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 310,8 g."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is driehonderdtien komma acht gram."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 75.633 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vijfenzeventigduizendzeshonderddrieëndertig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1963 met 18%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd drieënzestig met achttien procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 16 - 20 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zestien tot twintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 53.695 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal drieënvijftigduizendzeshonderdvijfennegentig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1955 met 36%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vijfenvijftig met zesendertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 431m."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is vierhonderdeenendertig meter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 6 - 13 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zes tot dertien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 673.583 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zeshonderddrieënzeventigduizendvijfhonderddrieëntachtig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 243.865 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal tweehonderddrieënveertigduizendachthonderdvijfenzestig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 39.302.099 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal negenendertig miljoen driehonderdtweeduizendnegenennegentig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 6% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zes procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 99.013.872 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal negenennegentig miljoen dertienduizendachthonderdtweeënzeventig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 1 °C."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is één graad Celsius."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 26-29 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zesentwintig tot negenentwintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 758.826 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zevenhonderdachtenvijftigduizendachthonderdzesentwintig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 81,8% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar eenentachtig komma acht procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 12 - 13 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen twaalf tot dertien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 3cm."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is drie centimeter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 7-11 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zeven tot elf bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 66 GB."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is zesenzestig gigabyte."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 26% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zesentwintig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 30-38 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen dertig tot achtendertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 11-21 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen elf tot eenentwintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1989 met 16,8%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd negenentachtig met zestien komma acht procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 29 - 35 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen negenentwintig tot vijfendertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 309°C."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is driehonderdnegen graden Celsius."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 330 km."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is driehonderddertig kilometer."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 71.859 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal eenenzeventigduizendachthonderdnegenenvijftig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 100% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar honderd procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 98 cm."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is achtennegentig centimeter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 9.685 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal negenduizendzeshonderdvijfentachtig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 6L."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is zes liter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 4km."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is vier kilometer."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 96.558 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zesennegentigduizendvijfhonderdachtenvijftig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 86.481 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zesentachtigduizendvierhonderdeenentachtig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 985.645 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal negenhonderdvijfentachtigduizendzeshonderdvijfenveertig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 24-31 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen vierentwintig tot eenendertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 16% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zestien procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2025 met 89%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendvijfentwintig met negenentachtig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 86% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zesentachtig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 329.492 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal driehonderdnegenentwintigduizendvierhonderdtweeënnegentig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2026 met 25%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendzesentwintig met vijfentwintig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 92% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar tweeënnegentig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 23–29 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen drieëntwintig tot negenentwintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 121 °C."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is honderdeenentwintig graden Celsius."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1982 met 51%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd tweeëntachtig met eenenvijftig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 946.721 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal negenhonderdzesenveertigduizendzevenhonderdeenentwintig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1977 met 83%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd zevenenzeventig met drieëntachtig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1973 met 99,4%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd drieënzeventig met negenennegentig komma vier procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 4–5 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen vier tot vijf bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 19,4 graden."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is negentien komma vier graden Celsius."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2034 met 19,5%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendvierendertig met negentien komma vijf procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 72,9% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar tweeënzeventig komma negen procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 4 - 10 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen vier tot tien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 31% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar eenendertig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 78% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar achtenzeventig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 44.408 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vierenveertigduizendvierhonderdacht studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 21% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar eenentwintig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 14,2 GB."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is veertien komma twee gigabyte."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 48% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar achtenveertig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 83% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar drieëntachtig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 23% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar drieëntwintig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 15.932 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vijftienduizendnegenhonderdtweeëndertig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1957 met 91%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd zevenenvijftig met eenennegentig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1985 met 10,4%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vijfentachtig met tien komma vier procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 11-19 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen elf tot negentien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1995 met 30%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vijfennegentig met dertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 4715 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vierduizendzevenhonderdvijftien studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 55% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar vijfenvijftig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 359.613 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal driehonderdnegenenvijftigduizendzeshonderddertien studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 17.111 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zeventienduizendhonderdelf studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 1,2 GB."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is één komma twee gigabyte."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2034 met 56%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendvierendertig met zesenvijftig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 90% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar negentig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 130 cm."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is honderddertig centimeter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 85.076 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vijfentachtigduizendzesenzeventig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2023 met 13,9%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizenddrieëntwintig met dertien komma negen procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 69% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar negenenzestig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 3.387.029 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal drie miljoen driehonderdzevenentachtigduizendnegenentwintig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 47,3% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zevenenveertig komma drie procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2031 met 44%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendeenendertig met vierenveertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 75.978 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal vijfenzeventigduizendnegenhonderdachtenzeventig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2011 met 45%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendelf met vijfenveertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 6–12 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zes tot twaalf bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 6985 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zesduizendnegenhonderdvijfentachtig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 30-33 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen dertig tot drieëndertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 39% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar negenendertig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1994 met 85%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vierennegentig met vijfentachtig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 1.435 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal duizendvierhonderdvijfendertig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2019 met 17%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendnegentien met zeventien procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 49% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar negenenveertig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2027 met 27%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendzevenentwintig met zevenentwintig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1966 met 64,3%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd zesenzestig met vierenzestig komma drie procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2027 met 38%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendzevenentwintig met achtendertig procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 2026 met 64,1%."}, {"role": "assistant", "content": "De omzet steeg in tweeduizendzesentwintig met vierenzestig komma een procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 5,4kg."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is vijf komma vier kilo."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 9 - 16 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen negen tot zestien bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 26–36 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen zesentwintig tot zesendertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 22% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar tweeëntwintig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 3-8 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen drie tot acht bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 20 - 21 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen twintig tot eenentwintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 321kg."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is driehonderdeenentwintig kilo."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1987 met 45,6%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd zevenentachtig met vijfenveertig komma zes procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 13.158 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal dertienduizendhonderdachtenvijftig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1967 met 9,8%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd zevenenzestig met negen komma acht procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 29 - 39 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen negenentwintig tot negenendertig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 26.805 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zesentwintigduizendachthonderdvijf studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 51.408.499 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal eenenvijftig miljoen vierhonderdachtduizendvierhonderdnegenennegentig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 701.003 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zevenhonderdéénduizenddrie studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De monteur komt morgen tussen 20-21 bij u langs om de verbinding te controleren."}, {"role": "assistant", "content": "De monteur komt morgen tussen twintig tot eenentwintig bij u langs om de verbinding te controleren."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1958 met 58,6%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd achtenvijftig met achtenvijftig komma zes procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 4 km."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is vier kilometer."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1950 met 95,1%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd vijftig met vijfennegentig komma een procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Uw rentepercentage is gewijzigd naar 70% per jaar."}, {"role": "assistant", "content": "Uw rentepercentage is gewijzigd naar zeventig procent per jaar."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De omzet steeg in 1963 met 93,6%."}, {"role": "assistant", "content": "De omzet steeg in negentien honderd drieënzestig met drieënnegentig komma zes procent."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 358.705 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal driehonderdachtenvijftigduizendzevenhonderdvijf studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 13.154 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal dertienduizendhonderdvierenvijftig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De maximale afmeting voor dit type zending is 3ml."}, {"role": "assistant", "content": "De maximale afmeting voor dit type zending is drie milliliter."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn in totaal 7.662 studenten ingeschreven aan onze universiteit."}, {"role": "assistant", "content": "Er zijn in totaal zevenduizendzeshonderdtweeënzestig studenten ingeschreven aan onze universiteit."}], "lang": "nl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze MH370 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze MH trzy siedem zero zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 62 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu sześćdziesiąt dwa dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 80,17zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiemdziesiąt złotych siedemnaście groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze AK6120 zostanie dostarczona w ciągu 975 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu dziewięćset siedemdziesiąt pięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze X-12340567 zostało opłacone kwotą 28 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostało opłacone kwotą dwadzieścia osiem złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze WXY 1234 zostało opłacone kwotą 671,90 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze WXY jeden dwa trzy cztery zostało opłacone kwotą sześćset siedemdziesiąt jeden złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 46 654,90zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi czterdzieści sześć tysięcy sześćset pięćdziesiąt cztery złote dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze B2-14 zostało opłacone kwotą 56 153,50 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze B dwa jeden cztery zostało opłacone kwotą pięćdziesiąt sześć tysięcy sto pięćdziesiąt trzy złote pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze PO 55219 zostało opłacone kwotą 85 589 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze PO pięć pięć dwa jeden dziewięć zostało opłacone kwotą osiemdziesiąt pięć tysięcy pięćset osiemdziesiąt dziewięć złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze A12 zostanie dostarczona w ciągu 878 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze A jeden dwa zostanie dostarczona w ciągu osiemset siedemdziesiąt osiem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 975,61zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dziewięćset siedemdziesiąt pięć złotych sześćdziesiąt jeden groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą PLN 24 913,90."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą dwadzieścia cztery tysiące dziewięćset trzynaście złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 9 033,72 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dziewięć tysięcy trzydzieści trzy złote siedemdziesiąt dwa grosze, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 717,90 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedemset siedemnaście złotych dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze ABC1234 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze ABC jeden dwa trzy cztery zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 59zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi pięćdziesiąt dziewięć złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze B2-14 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze B dwa jeden cztery zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 333,80, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi trzysta trzydzieści trzy złote osiemdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze TK-88213 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze TK osiem osiem dwa jeden trzy zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 81 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiemdziesiąt jeden złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze INV0472 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze INV zero cztery siedem dwa zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 61, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sześćdziesiąt jeden złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 9 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu dziewięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze X-12340567 zostało opłacone kwotą 80 626,90 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostało opłacone kwotą osiemdziesiąt tysięcy sześćset dwadzieścia sześć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze INV0472 zostanie dostarczona w ciągu 91 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu dziewięćdziesiąt jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 430,29, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi czterysta trzydzieści złotych dwadzieścia dziewięć groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze A12 zostało opłacone kwotą 43 434,82 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze A jeden dwa zostało opłacone kwotą czterdzieści trzy tysiące czterysta trzydzieści cztery złote osiemdziesiąt dwa grosze."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze INV0472 zostanie dostarczona w ciągu 51 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu pięćdziesiąt jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze MH370 zostanie dostarczona w ciągu 9 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze MH trzy siedem zero zostanie dostarczona w ciągu dziewięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 105,90, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sto pięć złotych dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 69,50, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sześćdziesiąt dziewięć złotych pięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze ABC1234 zostanie dostarczona w ciągu 3 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze ABC jeden dwa trzy cztery zostanie dostarczona w ciągu trzy dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze B2-14 zostało opłacone kwotą 34zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze B dwa jeden cztery zostało opłacone kwotą trzydzieści cztery złote."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze AK6120 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze AK sześć jeden dwa zero zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze QZ8501 zostało opłacone kwotą 66 753 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze QZ osiem pięć zero jeden zostało opłacone kwotą sześćdziesiąt sześć tysięcy siedemset pięćdziesiąt trzy złote."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze X-12340567 zostało opłacone kwotą 90zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostało opłacone kwotą dziewięćdziesiąt złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze INV0472 zostanie dostarczona w ciągu 1 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze PO 55219 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze PO pięć pięć dwa jeden dziewięć zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze X-12340567 zostało opłacone kwotą PLN 20 894,90."}, {"role": "assistant", "content": "Twoje zamówienie o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostało opłacone kwotą dwadzieścia tysięcy osiemset dziewięćdziesiąt cztery złote dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze QZ8501 zostanie dostarczona w ciągu 47 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze QZ osiem pięć zero jeden zostanie dostarczona w ciągu czterdzieści siedem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze ABC1234 zostało opłacone kwotą 454,50zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze ABC jeden dwa trzy cztery zostało opłacone kwotą czterysta pięćdziesiąt cztery złote pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze X-12340567 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 51 651zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi pięćdziesiąt jeden tysięcy sześćset pięćdziesiąt jeden złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze PO 55219 zostanie dostarczona w ciągu 2 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze PO pięć pięć dwa jeden dziewięć zostanie dostarczona w ciągu dwa dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 87 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiemdziesiąt siedem złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 938,50zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dziewięćset trzydzieści osiem złotych pięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze PO 55219 zostanie dostarczona w ciągu 6 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze PO pięć pięć dwa jeden dziewięć zostanie dostarczona w ciągu sześć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą PLN 799,90."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą siedemset dziewięćdziesiąt dziewięć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 472,50zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi czterysta siedemdziesiąt dwa złote pięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze REF-2024-0113 zostanie dostarczona w ciągu 167 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze REF dwa zero dwa cztery zero jeden jeden trzy zostanie dostarczona w ciągu sto sześćdziesiąt siedem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 37 177zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi trzydzieści siedem tysięcy sto siedemdziesiąt siedem złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze B2-14 zostanie dostarczona w ciągu 3 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze B dwa jeden cztery zostanie dostarczona w ciągu trzy dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze AK6120 zostanie dostarczona w ciągu 29 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu dwadzieścia dziewięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze ABC1234 zostało opłacone kwotą PLN 85,90."}, {"role": "assistant", "content": "Twoje zamówienie o numerze ABC jeden dwa trzy cztery zostało opłacone kwotą osiemdziesiąt pięć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą PLN 69."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą sześćdziesiąt dziewięć złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze AK6120 zostanie dostarczona w ciągu 1 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze A12 zostało opłacone kwotą 678,82zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze A jeden dwa zostało opłacone kwotą sześćset siedemdziesiąt osiem złotych osiemdziesiąt dwa grosze."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 4, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi cztery złote, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze INV0472 zostanie dostarczona w ciągu 31 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu trzydzieści jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 860zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiemset sześćdziesiąt złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 7 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedem złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze A12 zostało opłacone kwotą 27,50 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze A jeden dwa zostało opłacone kwotą dwadzieścia siedem złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze TK-88213 zostanie dostarczona w ciągu 19 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze TK osiem osiem dwa jeden trzy zostanie dostarczona w ciągu dziewiętnaście dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 49 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu czterdzieści dziewięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze A12 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze A jeden dwa zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze B2-14 zostało opłacone kwotą 17,50zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze B dwa jeden cztery zostało opłacone kwotą siedemnaście złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 147,20zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sto czterdzieści siedem złotych dwadzieścia groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 67 311 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sześćdziesiąt siedem tysięcy trzysta jedenaście złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 75 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu siedemdziesiąt pięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze TK-88213 zostało opłacone kwotą PLN 50 671,50."}, {"role": "assistant", "content": "Twoje zamówienie o numerze TK osiem osiem dwa jeden trzy zostało opłacone kwotą pięćdziesiąt tysięcy sześćset siedemdziesiąt jeden złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 14 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu czternaście dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze B2-14 zostało opłacone kwotą 37,90zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze B dwa jeden cztery zostało opłacone kwotą trzydzieści siedem złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą 31 645,90zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą trzydzieści jeden tysięcy sześćset czterdzieści pięć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze TK-88213 zostanie dostarczona w ciągu 265 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze TK osiem osiem dwa jeden trzy zostanie dostarczona w ciągu dwieście sześćdziesiąt pięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze PO 55219 zostało opłacone kwotą 649,90zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze PO pięć pięć dwa jeden dziewięć zostało opłacone kwotą sześćset czterdzieści dziewięć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze QZ8501 zostanie dostarczona w ciągu 57 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze QZ osiem pięć zero jeden zostanie dostarczona w ciągu pięćdziesiąt siedem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze QZ8501 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze QZ osiem pięć zero jeden zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze QZ8501 zostanie dostarczona w ciągu 413 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze QZ osiem pięć zero jeden zostanie dostarczona w ciągu czterysta trzynaście dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze PO 55219 zostało opłacone kwotą 607zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze PO pięć pięć dwa jeden dziewięć zostało opłacone kwotą sześćset siedem złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Zgłoszenie o numerze REF-2024-0113 zostało przekazane do działu technicznego."}, {"role": "assistant", "content": "Zgłoszenie o numerze REF dwa zero dwa cztery zero jeden jeden trzy zostało przekazane do działu technicznego."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 87zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiemdziesiąt siedem złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze AK6120 zostanie dostarczona w ciągu 156 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu sto pięćdziesiąt sześć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze INV0472 zostanie dostarczona w ciągu 97 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu dziewięćdziesiąt siedem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze X-12340567 zostanie dostarczona w ciągu 1 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze X jeden dwa trzy cztery zero pięć sześć siedem zostanie dostarczona w ciągu jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze AK6120 zostało opłacone kwotą PLN 79,50."}, {"role": "assistant", "content": "Twoje zamówienie o numerze AK sześć jeden dwa zero zostało opłacone kwotą siedemdziesiąt dziewięć złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 791,35, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedemset dziewięćdziesiąt jeden złotych trzydzieści pięć groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze B2-14 zostanie dostarczona w ciągu 508 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze B dwa jeden cztery zostanie dostarczona w ciągu pięćset osiem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 51 163, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi pięćdziesiąt jeden tysięcy sto sześćdziesiąt trzy złote, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze B2-14 zostanie dostarczona w ciągu 30 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze B dwa jeden cztery zostanie dostarczona w ciągu trzydzieści dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze REF-2024-0113 zostało opłacone kwotą 69 773,41zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze REF dwa zero dwa cztery zero jeden jeden trzy zostało opłacone kwotą sześćdziesiąt dziewięć tysięcy siedemset siedemdziesiąt trzy złote czterdzieści jeden groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 3 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu trzy dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 25 228,50, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dwadzieścia pięć tysięcy dwieście dwadzieścia osiem złotych pięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą 52 132,90 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą pięćdziesiąt dwa tysiące sto trzydzieści dwa złote dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 19 881, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dziewiętnaście tysięcy osiemset osiemdziesiąt jeden złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 73zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedemdziesiąt trzy złote, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze A12 zostało opłacone kwotą 649 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze A jeden dwa zostało opłacone kwotą sześćset czterdzieści dziewięć złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą PLN 70."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą siedemdziesiąt złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze A12 zostanie dostarczona w ciągu 1 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze A jeden dwa zostanie dostarczona w ciągu jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze INV0472 zostało opłacone kwotą PLN 81 361."}, {"role": "assistant", "content": "Twoje zamówienie o numerze INV zero cztery siedem dwa zostało opłacone kwotą osiemdziesiąt jeden tysięcy trzysta sześćdziesiąt jeden złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze AK6120 zostanie dostarczona w ciągu 96 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu dziewięćdziesiąt sześć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze AK6120 zostało opłacone kwotą PLN 37 386,90."}, {"role": "assistant", "content": "Twoje zamówienie o numerze AK sześć jeden dwa zero zostało opłacone kwotą trzydzieści siedem tysięcy trzysta osiemdziesiąt sześć złotych dziewięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze TK-88213 zostanie dostarczona w ciągu 1 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze TK osiem osiem dwa jeden trzy zostanie dostarczona w ciągu jeden dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 75 948,50zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedemdziesiąt pięć tysięcy dziewięćset czterdzieści osiem złotych pięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze PO 55219 zostało opłacone kwotą PLN 499."}, {"role": "assistant", "content": "Twoje zamówienie o numerze PO pięć pięć dwa jeden dziewięć zostało opłacone kwotą czterysta dziewięćdziesiąt dziewięć złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 732,90zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi siedemset trzydzieści dwa złote dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 310,90, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi trzysta dziesięć złotych dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze ABC1234 zostało opłacone kwotą 62 487 zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze ABC jeden dwa trzy cztery zostało opłacone kwotą sześćdziesiąt dwa tysiące czterysta osiemdziesiąt siedem złotych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze TK-88213 zostało opłacone kwotą PLN 16 060,50."}, {"role": "assistant", "content": "Twoje zamówienie o numerze TK osiem osiem dwa jeden trzy zostało opłacone kwotą szesnaście tysięcy sześćdziesiąt złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje zamówienie o numerze AK6120 zostało opłacone kwotą 36 940,50zł."}, {"role": "assistant", "content": "Twoje zamówienie o numerze AK sześć jeden dwa zero zostało opłacone kwotą trzydzieści sześć tysięcy dziewięćset czterdzieści złotych pięćdziesiąt groszy."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 688,90 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sześćset osiemdziesiąt osiem złotych dziewięćdziesiąt groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze B2-14 zostanie dostarczona w ciągu 99 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze B dwa jeden cztery zostanie dostarczona w ciągu dziewięćdziesiąt dziewięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 8 592 zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi osiem tysięcy pięćset dziewięćdziesiąt dwa złote, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja przesyłka o numerze AK6120 zostanie dostarczona w ciągu 8 dni roboczych."}, {"role": "assistant", "content": "Twoja przesyłka o numerze AK sześć jeden dwa zero zostanie dostarczona w ciągu osiem dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze WXY 1234 zostanie dostarczona w ciągu 795 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze WXY jeden dwa trzy cztery zostanie dostarczona w ciągu siedemset dziewięćdziesiąt pięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi PLN 26 575, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi dwadzieścia sześć tysięcy pięćset siedemdziesiąt pięć złotych, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Koszt konsultacji wynosi 158,13zł, płatność można uregulować w recepcji."}, {"role": "assistant", "content": "Koszt konsultacji wynosi sto pięćdziesiąt osiem złotych trzynaście groszy, płatność można uregulować w recepcji."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przesyłka o numerze INV0472 zostanie dostarczona w ciągu 975 dni roboczych."}, {"role": "assistant", "content": "Przesyłka o numerze INV zero cztery siedem dwa zostanie dostarczona w ciągu dziewięćset siedemdziesiąt pięć dni roboczych."}], "lang": "pl", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 11-03-2006 pukul 17:30 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sebelas Mac dua ribu enam pukul tujuh belas tiga puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 4 November 2028 pukul 4:10 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke empat November dua ribu dua puluh lapan pukul empat sepuluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM735.50, tenure 3 tahun, interest 36% setahun."}, {"role": "assistant", "content": "Loan approve tujuh ratus tiga puluh lima ringgit lima puluh sen, tenure tiga tahun, interest tiga puluh enam peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 17 April pukul 13:20 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tujuh belas April pukul tiga belas dua puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 72,938.95, tenure 94 tahun, interest 82.9% setahun."}, {"role": "assistant", "content": "Loan approve tujuh puluh dua ribu sembilan ratus tiga puluh lapan ringgit sembilan puluh lima sen, tenure sembilan puluh empat tahun, interest lapan puluh dua perpuluhan sembilan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 19 Disember pukul 6:30 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sembilan belas Disember pukul enam tiga puluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 13,453.93, tenure 92 tahun, interest 66.7% setahun."}, {"role": "assistant", "content": "Loan approve tiga belas ribu empat ratus lima puluh tiga ringgit sembilan puluh tiga sen, tenure sembilan puluh dua tahun, interest enam puluh enam perpuluhan tujuh peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM88, tenure 506 tahun, interest 88% setahun."}, {"role": "assistant", "content": "Loan approve lapan puluh lapan ringgit, tenure lima ratus enam tahun, interest lapan puluh lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM285.50, tenure 710 tahun, interest 29.8% setahun."}, {"role": "assistant", "content": "Loan approve dua ratus lapan puluh lima ringgit lima puluh sen, tenure tujuh ratus sepuluh tahun, interest dua puluh sembilan perpuluhan lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 40,127, tenure 881 tahun, interest 92% setahun."}, {"role": "assistant", "content": "Loan approve empat puluh ribu seratus dua puluh tujuh ringgit, tenure lapan ratus lapan puluh satu tahun, interest sembilan puluh dua peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 26 Februari pukul 6.50 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh enam Februari pukul enam lima puluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 4 Jun 2010 pukul 2:00 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke empat Jun dua ribu sepuluh pukul dua petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 21,589.90, tenure 2 tahun, interest 94% setahun."}, {"role": "assistant", "content": "Loan approve dua puluh satu ribu lima ratus lapan puluh sembilan ringgit sembilan puluh sen, tenure dua tahun, interest sembilan puluh empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM13.87, tenure 3 tahun, interest 41% setahun."}, {"role": "assistant", "content": "Loan approve tiga belas ringgit lapan puluh tujuh sen, tenure tiga tahun, interest empat puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 1 Januari pukul 4.45 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke satu Januari pukul empat empat puluh lima petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM12,424, tenure 1 tahun, interest 48% setahun."}, {"role": "assistant", "content": "Loan approve dua belas ribu empat ratus dua puluh empat ringgit, tenure satu tahun, interest empat puluh lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 17 November pukul 11:40 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tujuh belas November pukul sebelas empat puluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 15/8/2014 pukul 8:10 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke lima belas Ogos dua ribu empat belas pukul lapan sepuluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 25-08-1996 pukul 6 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh lima Ogos seribu sembilan ratus sembilan puluh enam pukul enam pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM43.50, tenure 320 tahun, interest 87.8% setahun."}, {"role": "assistant", "content": "Loan approve empat puluh tiga ringgit lima puluh sen, tenure tiga ratus dua puluh tahun, interest lapan puluh tujuh perpuluhan lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 23-07-2004 pukul 12:00 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh tiga Julai dua ribu empat pukul dua belas petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 11-02-2009 pukul 1:45 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sebelas Februari dua ribu sembilan pukul satu empat puluh lima malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 14 November 2025 pukul 0:00 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke empat belas November dua ribu dua puluh lima pukul sifar kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM58.90, tenure 1 tahun, interest 82% setahun."}, {"role": "assistant", "content": "Loan approve lima puluh lapan ringgit sembilan puluh sen, tenure satu tahun, interest lapan puluh dua peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM50.50, tenure 82 tahun, interest 21% setahun."}, {"role": "assistant", "content": "Loan approve lima puluh ringgit lima puluh sen, tenure lapan puluh dua tahun, interest dua puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 24-08-2031 pukul 1.50 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh empat Ogos dua ribu tiga puluh satu pukul satu lima puluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 24/12/2000 pukul 22:40 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh empat Disember dua ribu pukul dua puluh dua empat puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 21/4/2006 pukul 11:30 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh satu April dua ribu enam pukul sebelas tiga puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 22,392.74, tenure 110 tahun, interest 54% setahun."}, {"role": "assistant", "content": "Loan approve dua puluh dua ribu tiga ratus sembilan puluh dua ringgit tujuh puluh empat sen, tenure seratus sepuluh tahun, interest lima puluh empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 20-07-2029 pukul 3 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh Julai dua ribu dua puluh sembilan pukul tiga petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM885.90, tenure 10 tahun, interest 24% setahun."}, {"role": "assistant", "content": "Loan approve lapan ratus lapan puluh lima ringgit sembilan puluh sen, tenure sepuluh tahun, interest dua puluh empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 26 Januari 2008 pukul 6.00 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh enam Januari dua ribu lapan pukul enam petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM805.90, tenure 945 tahun, interest 19% setahun."}, {"role": "assistant", "content": "Loan approve lapan ratus lima ringgit sembilan puluh sen, tenure sembilan ratus empat puluh lima tahun, interest sembilan belas peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 676.90, tenure 5 tahun, interest 57% setahun."}, {"role": "assistant", "content": "Loan approve enam ratus tujuh puluh enam ringgit sembilan puluh sen, tenure lima tahun, interest lima puluh tujuh peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 21-11-2008 pukul 1.20 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh satu November dua ribu lapan pukul satu dua puluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 2 Disember pukul 14:30 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua Disember pukul empat belas tiga puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 6/4/2015 pukul 1:50 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke enam April dua ribu lima belas pukul satu lima puluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 26 Januari 2025 pukul 1 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh enam Januari dua ribu dua puluh lima pukul satu malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM19,099.50, tenure 971 tahun, interest 41% setahun."}, {"role": "assistant", "content": "Loan approve sembilan belas ribu sembilan puluh sembilan ringgit lima puluh sen, tenure sembilan ratus tujuh puluh satu tahun, interest empat puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 13 Mei pukul 9 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tiga belas Mei pukul sembilan pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM58,477.50, tenure 65 tahun, interest 10% setahun."}, {"role": "assistant", "content": "Loan approve lima puluh lapan ribu empat ratus tujuh puluh tujuh ringgit lima puluh sen, tenure enam puluh lima tahun, interest sepuluh peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM7, tenure 1 tahun, interest 56.9% setahun."}, {"role": "assistant", "content": "Loan approve tujuh ringgit, tenure satu tahun, interest lima puluh enam perpuluhan sembilan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 90, tenure 911 tahun, interest 18.5% setahun."}, {"role": "assistant", "content": "Loan approve sembilan puluh ringgit, tenure sembilan ratus sebelas tahun, interest lapan belas perpuluhan lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 25 Jun 1996 pukul 6.30 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh lima Jun seribu sembilan ratus sembilan puluh enam pukul enam tiga puluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 5 Oktober 1998 pukul 5:10 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke lima Oktober seribu sembilan ratus sembilan puluh lapan pukul lima sepuluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 22-03-2013 pukul 2 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh dua Mac dua ribu tiga belas pukul dua petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM791, tenure 1 tahun, interest 53% setahun."}, {"role": "assistant", "content": "Loan approve tujuh ratus sembilan puluh satu ringgit, tenure satu tahun, interest lima puluh tiga peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM26,114.90, tenure 1 tahun, interest 79.5% setahun."}, {"role": "assistant", "content": "Loan approve dua puluh enam ribu seratus empat belas ringgit sembilan puluh sen, tenure satu tahun, interest tujuh puluh sembilan perpuluhan lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 788, tenure 658 tahun, interest 95.8% setahun."}, {"role": "assistant", "content": "Loan approve tujuh ratus lapan puluh lapan ringgit, tenure enam ratus lima puluh lapan tahun, interest sembilan puluh lima perpuluhan lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 18-11-2003 pukul 0:00 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke lapan belas November dua ribu tiga pukul sifar kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM186.50, tenure 92 tahun, interest 80.8% setahun."}, {"role": "assistant", "content": "Loan approve seratus lapan puluh enam ringgit lima puluh sen, tenure sembilan puluh dua tahun, interest lapan puluh perpuluhan lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 11 Januari 2021 pukul 16:05 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sebelas Januari dua ribu dua puluh satu pukul enam belas kosong lima kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM62,645, tenure 3 tahun, interest 37.7% setahun."}, {"role": "assistant", "content": "Loan approve enam puluh dua ribu enam ratus empat puluh lima ringgit, tenure tiga tahun, interest tiga puluh tujuh perpuluhan tujuh peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM12,364.50, tenure 175 tahun, interest 70.6% setahun."}, {"role": "assistant", "content": "Loan approve dua belas ribu tiga ratus enam puluh empat ringgit lima puluh sen, tenure seratus tujuh puluh lima tahun, interest tujuh puluh perpuluhan enam peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 13/6/2017 pukul 12:40 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tiga belas Jun dua ribu tujuh belas pukul dua belas empat puluh petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 15,441.50, tenure 2 tahun, interest 43.2% setahun."}, {"role": "assistant", "content": "Loan approve lima belas ribu empat ratus empat puluh satu ringgit lima puluh sen, tenure dua tahun, interest empat puluh tiga perpuluhan dua peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 27-12-2015 pukul 12 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh tujuh Disember dua ribu lima belas pukul dua belas malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 22-06-2018 pukul 13:20 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh dua Jun dua ribu lapan belas pukul tiga belas dua puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 947.50, tenure 83 tahun, interest 35% setahun."}, {"role": "assistant", "content": "Loan approve sembilan ratus empat puluh tujuh ringgit lima puluh sen, tenure lapan puluh tiga tahun, interest tiga puluh lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM897.15, tenure 33 tahun, interest 15.1% setahun."}, {"role": "assistant", "content": "Loan approve lapan ratus sembilan puluh tujuh ringgit lima belas sen, tenure tiga puluh tiga tahun, interest lima belas perpuluhan satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM424.50, tenure 36 tahun, interest 21% setahun."}, {"role": "assistant", "content": "Loan approve empat ratus dua puluh empat ringgit lima puluh sen, tenure tiga puluh enam tahun, interest dua puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 55,128.40, tenure 8 tahun, interest 1% setahun."}, {"role": "assistant", "content": "Loan approve lima puluh lima ribu seratus dua puluh lapan ringgit empat puluh sen, tenure lapan tahun, interest satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 83, tenure 91 tahun, interest 33.1% setahun."}, {"role": "assistant", "content": "Loan approve lapan puluh tiga ringgit, tenure sembilan puluh satu tahun, interest tiga puluh tiga perpuluhan satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 22/7/2017 pukul 7:10 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh dua Julai dua ribu tujuh belas pukul tujuh sepuluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM486, tenure 42 tahun, interest 74% setahun."}, {"role": "assistant", "content": "Loan approve empat ratus lapan puluh enam ringgit, tenure empat puluh dua tahun, interest tujuh puluh empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM447.04, tenure 361 tahun, interest 62% setahun."}, {"role": "assistant", "content": "Loan approve empat ratus empat puluh tujuh ringgit empat sen, tenure tiga ratus enam puluh satu tahun, interest enam puluh dua peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM75.50, tenure 508 tahun, interest 81% setahun."}, {"role": "assistant", "content": "Loan approve tujuh puluh lima ringgit lima puluh sen, tenure lima ratus lapan tahun, interest lapan puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM38, tenure 1 tahun, interest 74.5% setahun."}, {"role": "assistant", "content": "Loan approve tiga puluh lapan ringgit, tenure satu tahun, interest tujuh puluh empat perpuluhan lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 499, tenure 760 tahun, interest 30% setahun."}, {"role": "assistant", "content": "Loan approve empat ratus sembilan puluh sembilan ringgit, tenure tujuh ratus enam puluh tahun, interest tiga puluh peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 11 Mac 2003 pukul 22:00 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sebelas Mac dua ribu tiga pukul dua puluh dua kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 25-10-2010 pukul 4 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh lima Oktober dua ribu sepuluh pukul empat petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM16, tenure 1 tahun, interest 53% setahun."}, {"role": "assistant", "content": "Loan approve enam belas ringgit, tenure satu tahun, interest lima puluh tiga peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 675, tenure 7 tahun, interest 93% setahun."}, {"role": "assistant", "content": "Loan approve enam ratus tujuh puluh lima ringgit, tenure tujuh tahun, interest sembilan puluh tiga peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 17-12-2012 pukul 6:10 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tujuh belas Disember dua ribu dua belas pukul enam sepuluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 27 September pukul 19:30 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh tujuh September pukul sembilan belas tiga puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 4 Februari 2031 pukul 9:05 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke empat Februari dua ribu tiga puluh satu pukul sembilan kosong lima malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM574, tenure 63 tahun, interest 14% setahun."}, {"role": "assistant", "content": "Loan approve lima ratus tujuh puluh empat ringgit, tenure enam puluh tiga tahun, interest empat belas peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM41.24, tenure 95 tahun, interest 98.5% setahun."}, {"role": "assistant", "content": "Loan approve empat puluh satu ringgit dua puluh empat sen, tenure sembilan puluh lima tahun, interest sembilan puluh lapan perpuluhan lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 28 Jun 2032 pukul 5 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua puluh lapan Jun dua ribu tiga puluh dua pukul lima petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM847.66, tenure 68 tahun, interest 91% setahun."}, {"role": "assistant", "content": "Loan approve lapan ratus empat puluh tujuh ringgit enam puluh enam sen, tenure enam puluh lapan tahun, interest sembilan puluh satu peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 84.88, tenure 9 tahun, interest 17% setahun."}, {"role": "assistant", "content": "Loan approve lapan puluh empat ringgit lapan puluh lapan sen, tenure sembilan tahun, interest tujuh belas peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 949.08, tenure 2 tahun, interest 66.4% setahun."}, {"role": "assistant", "content": "Loan approve sembilan ratus empat puluh sembilan ringgit lapan sen, tenure dua tahun, interest enam puluh enam perpuluhan empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 2 Februari pukul 2:20 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua Februari pukul dua dua puluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 1/7/2005 pukul 12 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke satu Julai dua ribu lima pukul dua belas malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 27,960.50, tenure 7 tahun, interest 8% setahun."}, {"role": "assistant", "content": "Loan approve dua puluh tujuh ribu sembilan ratus enam puluh ringgit lima puluh sen, tenure tujuh tahun, interest lapan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM42, tenure 27 tahun, interest 57.4% setahun."}, {"role": "assistant", "content": "Loan approve empat puluh dua ringgit, tenure dua puluh tujuh tahun, interest lima puluh tujuh perpuluhan empat peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 02-12-2004 pukul 8:40 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke dua Disember dua ribu empat pukul lapan empat puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 13 Disember 1997 pukul 13:30 kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tiga belas Disember seribu sembilan ratus sembilan puluh tujuh pukul tiga belas tiga puluh kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 09-09-2011 pukul 11 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sembilan September dua ribu sebelas pukul sebelas malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 13 Februari 1995 pukul 1:15 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tiga belas Februari seribu sembilan ratus sembilan puluh lima pukul satu lima belas malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 231, tenure 460 tahun, interest 89% setahun."}, {"role": "assistant", "content": "Loan approve dua ratus tiga puluh satu ringgit, tenure empat ratus enam puluh tahun, interest lapan puluh sembilan peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 92,204.61, tenure 4 tahun, interest 85% setahun."}, {"role": "assistant", "content": "Loan approve sembilan puluh dua ribu dua ratus empat ringgit enam puluh satu sen, tenure empat tahun, interest lapan puluh lima peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 9/10/2003 pukul 11.10 malam kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke sembilan Oktober dua ribu tiga pukul sebelas sepuluh malam kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 14-07-2028 pukul 8:40 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke empat belas Julai dua ribu dua puluh lapan pukul lapan empat puluh pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 7/1/2008 pukul 8.05 pagi kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke tujuh Januari dua ribu lapan pukul lapan kosong lima pagi kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 6/11/1999 pukul 3:10 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke enam November seribu sembilan ratus sembilan puluh sembilan pukul tiga sepuluh petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM42.09, tenure 45 tahun, interest 82% setahun."}, {"role": "assistant", "content": "Loan approve empat puluh dua ringgit sembilan sen, tenure empat puluh lima tahun, interest lapan puluh dua peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Meeting kita reschedule ke 18 Disember 2007 pukul 3.45 petang kat office."}, {"role": "assistant", "content": "Meeting kita reschedule ke lapan belas Disember dua ribu tujuh pukul tiga empat puluh lima petang kat office."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Malay with English mixed in. Malaysian Malay-English code-switching (Bahasa rojak). RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Malay part is spoken in Malay, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Loan approve RM 124, tenure 346 tahun, interest 16% setahun."}, {"role": "assistant", "content": "Loan approve seratus dua puluh empat ringgit, tenure tiga ratus empat puluh enam tahun, interest enam belas peratus setahun."}], "lang": "ms-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM46,947, tunggu 22-29 working days."}, {"role": "assistant", "content": "We already refund forty-six thousand nine hundred forty-seven ringgit, tunggu twenty-two to twenty-nine working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 633.50, tunggu 23-28 working days."}, {"role": "assistant", "content": "We already refund six hundred thirty-three US dollars fifty cents, tunggu twenty-three to twenty-eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $69.90, tunggu 30-33 working days."}, {"role": "assistant", "content": "We already refund sixty-nine dollars ninety cents, tunggu thirty to thirty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before April 15, 2010 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before April fifteenth, twenty ten to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM59,752.90, tunggu 24 - 33 working days."}, {"role": "assistant", "content": "We already refund fifty-nine thousand seven hundred fifty-two ringgit ninety sen, tunggu twenty-four to thirty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 15 October to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifteenth of October to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 23/4/2013 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-third of April twenty thirteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 59,480.28, tunggu 20 - 23 working days."}, {"role": "assistant", "content": "We already refund fifty-nine thousand four hundred eighty US dollars twenty-eight cents, tunggu twenty to twenty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 201.50, tunggu 23 - 24 working days."}, {"role": "assistant", "content": "We already refund two hundred one ringgit fifty sen, tunggu twenty-three to twenty-four working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM678, tunggu 5-13 working days."}, {"role": "assistant", "content": "We already refund six hundred seventy-eight ringgit, tunggu five to thirteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 12/12/2032 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twelfth of December twenty thirty-two to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 21 June 2015 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-first of June twenty fifteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 20-04-1996 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twentieth of April nineteen ninety-six to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 15-02-1998 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifteenth of February nineteen ninety-eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 477.90, tunggu 27-32 working days."}, {"role": "assistant", "content": "We already refund four hundred seventy-seven ringgit ninety sen, tunggu twenty-seven to thirty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 1, tunggu 8 - 9 working days."}, {"role": "assistant", "content": "We already refund one US dollars, tunggu eight to nine working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before February 5, 2027 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before February fifth, twenty twenty-seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $61.50, tunggu 2 - 4 working days."}, {"role": "assistant", "content": "We already refund sixty-one dollars fifty cents, tunggu two to four working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 920, tunggu 13-16 working days."}, {"role": "assistant", "content": "We already refund nine hundred twenty ringgit, tunggu thirteen to sixteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 61.01, tunggu 2-4 working days."}, {"role": "assistant", "content": "We already refund sixty-one US dollars one cents, tunggu two to four working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 68,806.90, tunggu 29–30 working days."}, {"role": "assistant", "content": "We already refund sixty-eight thousand eight hundred six US dollars ninety cents, tunggu twenty-nine to thirty working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 36.50, tunggu 22 - 25 working days."}, {"role": "assistant", "content": "We already refund thirty-six US dollars fifty cents, tunggu twenty-two to twenty-five working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24-09-2032 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of September twenty thirty-two to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 11 April 2026 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eleventh of April twenty twenty-six to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $428.17, tunggu 12-22 working days."}, {"role": "assistant", "content": "We already refund four hundred twenty-eight dollars seventeen cents, tunggu twelve to twenty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM20, tunggu 25-32 working days."}, {"role": "assistant", "content": "We already refund twenty ringgit, tunggu twenty-five to thirty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before December 4, 2028 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before December fourth, twenty twenty-eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 18 February 2001 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eighteenth of February two thousand and one to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24-05-1999 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of May nineteen ninety-nine to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before January 24, 2028 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before January twenty-fourth, twenty twenty-eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 5 November to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifth of November to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before July 14, 2007 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before July fourteenth, two thousand and seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 18.50, tunggu 3 - 4 working days."}, {"role": "assistant", "content": "We already refund eighteen US dollars fifty cents, tunggu three to four working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24 October 2002 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of October two thousand and two to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 07-07-1995 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the seventh of July nineteen ninety-five to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 333.57, tunggu 20 - 23 working days."}, {"role": "assistant", "content": "We already refund three hundred thirty-three US dollars fifty-seven cents, tunggu twenty to twenty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 76,308, tunggu 12–18 working days."}, {"role": "assistant", "content": "We already refund seventy-six thousand three hundred eight ringgit, tunggu twelve to eighteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 08-10-2008 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eighth of October two thousand and eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 9/5/2027 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the ninth of May twenty twenty-seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM3,661, tunggu 24-32 working days."}, {"role": "assistant", "content": "We already refund three thousand six hundred sixty-one ringgit, tunggu twenty-four to thirty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 557, tunggu 1-6 working days."}, {"role": "assistant", "content": "We already refund five hundred fifty-seven ringgit, tunggu one to six working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 2/11/2025 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the second of November twenty twenty-five to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $38, tunggu 26–29 working days."}, {"role": "assistant", "content": "We already refund thirty-eight dollars, tunggu twenty-six to twenty-nine working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 6/12/2007 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the sixth of December two thousand and seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $17, tunggu 21 - 22 working days."}, {"role": "assistant", "content": "We already refund seventeen dollars, tunggu twenty-one to twenty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24 March to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of March to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $68.06, tunggu 6–11 working days."}, {"role": "assistant", "content": "We already refund sixty-eight dollars six cents, tunggu six to eleven working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 89,973, tunggu 1-8 working days."}, {"role": "assistant", "content": "We already refund eighty-nine thousand nine hundred seventy-three US dollars, tunggu one to eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 18/2/2029 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eighteenth of February twenty twenty-nine to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $64,144.90, tunggu 8 - 10 working days."}, {"role": "assistant", "content": "We already refund sixty-four thousand one hundred forty-four dollars ninety cents, tunggu eight to ten working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 371.52, tunggu 29–32 working days."}, {"role": "assistant", "content": "We already refund three hundred seventy-one ringgit fifty-two sen, tunggu twenty-nine to thirty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM368, tunggu 8–17 working days."}, {"role": "assistant", "content": "We already refund three hundred sixty-eight ringgit, tunggu eight to seventeen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before May 24, 2027 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before May twenty-fourth, twenty twenty-seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 8 May to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eighth of May to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 8 June 1998 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the eighth of June nineteen ninety-eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before September 21, 2020 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before September twenty-first, twenty twenty to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 211, tunggu 14 - 23 working days."}, {"role": "assistant", "content": "We already refund two hundred eleven US dollars, tunggu fourteen to twenty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM192, tunggu 2-5 working days."}, {"role": "assistant", "content": "We already refund one hundred ninety-two ringgit, tunggu two to five working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 25 May to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fifth of May to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 11,103.90, tunggu 26–29 working days."}, {"role": "assistant", "content": "We already refund eleven thousand one hundred three ringgit ninety sen, tunggu twenty-six to twenty-nine working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 13 October to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the thirteenth of October to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 85.50, tunggu 24–30 working days."}, {"role": "assistant", "content": "We already refund eighty-five ringgit fifty sen, tunggu twenty-four to thirty working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before April 7, 2018 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before April seventh, twenty eighteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 32.90, tunggu 14–21 working days."}, {"role": "assistant", "content": "We already refund thirty-two US dollars ninety cents, tunggu fourteen to twenty-one working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 612.50, tunggu 17-19 working days."}, {"role": "assistant", "content": "We already refund six hundred twelve US dollars fifty cents, tunggu seventeen to nineteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 19 October to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the nineteenth of October to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 20 December 2006 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twentieth of December two thousand and six to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 14 May to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fourteenth of May to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 400.50, tunggu 26-35 working days."}, {"role": "assistant", "content": "We already refund four hundred ringgit fifty sen, tunggu twenty-six to thirty-five working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 21-12-2007 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-first of December two thousand and seven to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $57,559, tunggu 27 - 32 working days."}, {"role": "assistant", "content": "We already refund fifty-seven thousand five hundred fifty-nine dollars, tunggu twenty-seven to thirty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 10 December to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the tenth of December to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $73,436, tunggu 28-38 working days."}, {"role": "assistant", "content": "We already refund seventy-three thousand four hundred thirty-six dollars, tunggu twenty-eight to thirty-eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before October 20, 2023 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before October twentieth, twenty twenty-three to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 22-08-2031 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-second of August twenty thirty-one to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM65,620.90, tunggu 11-14 working days."}, {"role": "assistant", "content": "We already refund sixty-five thousand six hundred twenty ringgit ninety sen, tunggu eleven to fourteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 732.50, tunggu 28-36 working days."}, {"role": "assistant", "content": "We already refund seven hundred thirty-two ringgit fifty sen, tunggu twenty-eight to thirty-six working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 84.90, tunggu 23-25 working days."}, {"role": "assistant", "content": "We already refund eighty-four ringgit ninety sen, tunggu twenty-three to twenty-five working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 15 April to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifteenth of April to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 5 July 1996 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifth of July nineteen ninety-six to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24 July 2013 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of July twenty thirteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before October 20, 2003 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before October twentieth, two thousand and three to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 50,121.90, tunggu 24-28 working days."}, {"role": "assistant", "content": "We already refund fifty thousand one hundred twenty-one US dollars ninety cents, tunggu twenty-four to twenty-eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 26/8/2005 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-sixth of August two thousand and five to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 29,501.90, tunggu 19–28 working days."}, {"role": "assistant", "content": "We already refund twenty-nine thousand five hundred one US dollars ninety cents, tunggu nineteen to twenty-eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 57,029.90, tunggu 13–19 working days."}, {"role": "assistant", "content": "We already refund fifty-seven thousand twenty-nine ringgit ninety sen, tunggu thirteen to nineteen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 98,108.50, tunggu 25 - 34 working days."}, {"role": "assistant", "content": "We already refund ninety-eight thousand one hundred eight ringgit fifty sen, tunggu twenty-five to thirty-four working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 56.41, tunggu 23 - 33 working days."}, {"role": "assistant", "content": "We already refund fifty-six ringgit forty-one sen, tunggu twenty-three to thirty-three working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 36, tunggu 13–22 working days."}, {"role": "assistant", "content": "We already refund thirty-six ringgit, tunggu thirteen to twenty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 7 December to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the seventh of December to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 27/7/2025 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-seventh of July twenty twenty-five to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 27/8/2028 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-seventh of August twenty twenty-eight to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 61, tunggu 16-17 working days."}, {"role": "assistant", "content": "We already refund sixty-one US dollars, tunggu sixteen to seventeen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 24 August to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twenty-fourth of August to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $229.50, tunggu 7-11 working days."}, {"role": "assistant", "content": "We already refund two hundred twenty-nine dollars fifty cents, tunggu seven to eleven working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund USD 497.90, tunggu 25–28 working days."}, {"role": "assistant", "content": "We already refund four hundred ninety-seven US dollars ninety cents, tunggu twenty-five to twenty-eight working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $91.95, tunggu 24 - 26 working days."}, {"role": "assistant", "content": "We already refund ninety-one dollars ninety-five cents, tunggu twenty-four to twenty-six working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 07-10-2020 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the seventh of October twenty twenty to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 2 March to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the second of March to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $93.50, tunggu 6-11 working days."}, {"role": "assistant", "content": "We already refund ninety-three dollars fifty cents, tunggu six to eleven working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 605, tunggu 15-21 working days."}, {"role": "assistant", "content": "We already refund six hundred five ringgit, tunggu fifteen to twenty-one working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 05-08-2010 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the fifth of August twenty ten to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $327, tunggu 16-17 working days."}, {"role": "assistant", "content": "We already refund three hundred twenty-seven dollars, tunggu sixteen to seventeen working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 3,395.50, tunggu 23 - 30 working days."}, {"role": "assistant", "content": "We already refund three thousand three hundred ninety-five ringgit fifty sen, tunggu twenty-three to thirty working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 7 August 2018 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the seventh of August twenty eighteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please bayar before 12 August 2015 to avoid late charges."}, {"role": "assistant", "content": "Please bayar before the twelfth of August twenty fifteen to avoid late charges."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund $575.90, tunggu 30 - 35 working days."}, {"role": "assistant", "content": "We already refund five hundred seventy-five dollars ninety cents, tunggu thirty to thirty-five working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly English with Malay mixed in. Malaysian English with Malay mixed in. RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the English part is spoken in English, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We already refund RM 729.50, tunggu 21-22 working days."}, {"role": "assistant", "content": "We already refund seven hundred twenty-nine ringgit fifty sen, tunggu twenty-one to twenty-two working days."}], "lang": "en-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有131,814积分，可以redeem RM727.50的voucher。"}, {"role": "assistant", "content": "您有十三万一千八百一十四积分，可以redeem 七百二十七令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午12:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午十二点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有96,885积分，可以redeem RM 29,681的voucher。"}, {"role": "assistant", "content": "您有九万六千八百八十五积分，可以redeem 二万九千六百八十一令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午4点15分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午四点十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午8点40分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午八点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有94,823,834积分，可以redeem RM15,324的voucher。"}, {"role": "assistant", "content": "您有九千四百八十二万三千八百三十四积分，可以redeem 一万五千三百二十四令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有12,393积分，可以redeem RM 55.90的voucher。"}, {"role": "assistant", "content": "您有一万二千三百九十三积分，可以redeem 五十五令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有30,063,619积分，可以redeem RM31.50的voucher。"}, {"role": "assistant", "content": "您有三千零六万三千六百一十九积分，可以redeem 三十一令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有43,908,379积分，可以redeem RM 49.59的voucher。"}, {"role": "assistant", "content": "您有四千三百九十万八千三百七十九积分，可以redeem 四十九令吉五十九仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有30,605积分，可以redeem RM177的voucher。"}, {"role": "assistant", "content": "您有三万零六百零五积分，可以redeem 一百七十七令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在8点之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在八点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午12:30之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午十二点三十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在19:40之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十九点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午7:30之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午七点三十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有496,340积分，可以redeem RM 31,353的voucher。"}, {"role": "assistant", "content": "您有四十九万六千三百四十积分，可以redeem 三万一千三百五十三令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有40,831,219积分，可以redeem RM4.80的voucher。"}, {"role": "assistant", "content": "您有四千零八十三万一千二百一十九积分，可以redeem 四令吉八十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在1点之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在一点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有24,114,397积分，可以redeem RM9.86的voucher。"}, {"role": "assistant", "content": "您有二千四百一十一万四千三百九十七积分，可以redeem 九令吉八十六仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有99,519,962积分，可以redeem RM733的voucher。"}, {"role": "assistant", "content": "您有九千九百五十一万九千九百六十二积分，可以redeem 七百三十三令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有10,298,257积分，可以redeem RM95,691.90的voucher。"}, {"role": "assistant", "content": "您有一千零二十九万八千二百五十七积分，可以redeem 九万五千六百九十一令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有30,849积分，可以redeem RM75的voucher。"}, {"role": "assistant", "content": "您有三万零八百四十九积分，可以redeem 七十五令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在10点之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有62,067积分，可以redeem RM747的voucher。"}, {"role": "assistant", "content": "您有六万二千零六十七积分，可以redeem 七百四十七令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在1:10之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在一点十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在6:20之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在六点二十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在19:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十九点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在5:15之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在五点十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午10:50之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午十点五十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午9点05分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午九点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有143,317积分，可以redeem RM 26.90的voucher。"}, {"role": "assistant", "content": "您有十四万三千三百一十七积分，可以redeem 二十六令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有86,586积分，可以redeem RM 314.14的voucher。"}, {"role": "assistant", "content": "您有八万六千五百八十六积分，可以redeem 三百一十四令吉十四仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午8点40分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午八点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午1点40分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午一点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有19,692,908积分，可以redeem RM 68,993的voucher。"}, {"role": "assistant", "content": "您有一千九百六十九万二千九百零八积分，可以redeem 六万八千九百九十三令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午8:45之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午八点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有60,344积分，可以redeem RM81.90的voucher。"}, {"role": "assistant", "content": "您有六万零三百四十四积分，可以redeem 八十一令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有11,765,136积分，可以redeem RM411.51的voucher。"}, {"role": "assistant", "content": "您有一千一百七十六万五千一百三十六积分，可以redeem 四百一十一令吉五十一仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有794,613积分，可以redeem RM 91,606.90的voucher。"}, {"role": "assistant", "content": "您有七十九万四千六百一十三积分，可以redeem 九万一千六百零六令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在4:15之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在四点十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有86,497积分，可以redeem RM48.05的voucher。"}, {"role": "assistant", "content": "您有八万六千四百九十七积分，可以redeem 四十八令吉五仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有65,282,052积分，可以redeem RM 13,039的voucher。"}, {"role": "assistant", "content": "您有六千五百二十八万二千零五十二积分，可以redeem 一万三千零三十九令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有316,794积分，可以redeem RM 696.50的voucher。"}, {"role": "assistant", "content": "您有三十一万六千七百九十四积分，可以redeem 六百九十六令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在2:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在两点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午6点40分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午六点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在5:45之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在五点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有7,553,093积分，可以redeem RM 71,600的voucher。"}, {"role": "assistant", "content": "您有七百五十五万三千零九十三积分，可以redeem 七万一千六百令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午1:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午一点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有504,070积分，可以redeem RM12,805的voucher。"}, {"role": "assistant", "content": "您有五十万四千零七十积分，可以redeem 一万二千八百零五令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午2:15之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午两点十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在5点之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在五点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午11:20之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午十一点二十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有94,786积分，可以redeem RM68,382的voucher。"}, {"role": "assistant", "content": "您有九万四千七百八十六积分，可以redeem 六万八千三百八十二令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在5:05之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在五点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午4:45之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午四点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有9723积分，可以redeem RM 134.90的voucher。"}, {"role": "assistant", "content": "您有九千七百二十三积分，可以redeem 一百三十四令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有9,300积分，可以redeem RM5,560的voucher。"}, {"role": "assistant", "content": "您有九千三百积分，可以redeem 五千五百六十令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有55,188积分，可以redeem RM 96的voucher。"}, {"role": "assistant", "content": "您有五万五千一百八十八积分，可以redeem 九十六令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有180,561积分，可以redeem RM24,473.50的voucher。"}, {"role": "assistant", "content": "您有十八万零五百六十一积分，可以redeem 二万四千四百七十三令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有81,388,035积分，可以redeem RM540的voucher。"}, {"role": "assistant", "content": "您有八千一百三十八万八千零三十五积分，可以redeem 五百四十令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午11:05之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午十一点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有3263积分，可以redeem RM 80.50的voucher。"}, {"role": "assistant", "content": "您有三千二百六十三积分，可以redeem 八十令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有54,065,679积分，可以redeem RM 639.90的voucher。"}, {"role": "assistant", "content": "您有五千四百零六万五千六百七十九积分，可以redeem 六百三十九令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有9816积分，可以redeem RM77的voucher。"}, {"role": "assistant", "content": "您有九千八百一十六积分，可以redeem 七十七令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在16:40之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十六点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午3点45分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午三点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有905,209积分，可以redeem RM 73,892的voucher。"}, {"role": "assistant", "content": "您有九十万五千二百零九积分，可以redeem 七万三千八百九十二令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有49,245积分，可以redeem RM 66,937的voucher。"}, {"role": "assistant", "content": "您有四万九千二百四十五积分，可以redeem 六万六千九百三十七令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有94,595,322积分，可以redeem RM747.50的voucher。"}, {"role": "assistant", "content": "您有九千四百五十九万五千三百二十二积分，可以redeem 七百四十七令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有238,804积分，可以redeem RM86.43的voucher。"}, {"role": "assistant", "content": "您有二十三万八千八百零四积分，可以redeem 八十六令吉四十三仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午3点05分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午三点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有5,591积分，可以redeem RM86.50的voucher。"}, {"role": "assistant", "content": "您有五千五百九十一积分，可以redeem 八十六令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有29,979,298积分，可以redeem RM 67,129的voucher。"}, {"role": "assistant", "content": "您有二千九百九十七万九千二百九十八积分，可以redeem 六万七千一百二十九令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在20:15之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在二十点十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有48,504,815积分，可以redeem RM 91.50的voucher。"}, {"role": "assistant", "content": "您有四千八百五十万四千八百一十五积分，可以redeem 九十一令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有92,019,181积分，可以redeem RM219.50的voucher。"}, {"role": "assistant", "content": "您有九千二百零一万九千一百八十一积分，可以redeem 二百一十九令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在21:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在二十一点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午3点45分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午三点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午2:00之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午两点之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有5043积分，可以redeem RM 6,615.34的voucher。"}, {"role": "assistant", "content": "您有五千零四十三积分，可以redeem 六千六百一十五令吉三十四仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有13,959积分，可以redeem RM95,057的voucher。"}, {"role": "assistant", "content": "您有一万三千九百五十九积分，可以redeem 九万五千零五十七令吉的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有5732积分，可以redeem RM150.90的voucher。"}, {"role": "assistant", "content": "您有五千七百三十二积分，可以redeem 一百五十令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有5,346积分，可以redeem RM 450.50的voucher。"}, {"role": "assistant", "content": "您有五千三百四十六积分，可以redeem 四百五十令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午8:10之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午八点十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午10:30之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午十点三十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午11:40之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午十一点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有8,960积分，可以redeem RM 249.50的voucher。"}, {"role": "assistant", "content": "您有八千九百六十积分，可以redeem 二百四十九令吉五十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午8:40之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午八点四十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在15:10之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十五点十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有4149积分，可以redeem RM 27.90的voucher。"}, {"role": "assistant", "content": "您有四千一百四十九积分，可以redeem 二十七令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有4654积分，可以redeem RM84,210.90的voucher。"}, {"role": "assistant", "content": "您有四千六百五十四积分，可以redeem 八万四千二百一十令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午11:05之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午十一点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您有974,987积分，可以redeem RM 88,797.90的voucher。"}, {"role": "assistant", "content": "您有九十七万四千九百八十七积分，可以redeem 八万八千七百九十七令吉九十仙的voucher。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在下午1点10分之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在下午一点十分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在上午3:45之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在上午三点四十五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with English mixed in. Malaysian Mandarin with English mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在17:05之前完成payment，谢谢您。"}, {"role": "assistant", "content": "请在十七点零五分之前完成payment，谢谢您。"}], "lang": "zh-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM502，可以在www.posta.it付款。"}, {"role": "assistant", "content": "Yuran是五百零二令吉，可以在w w w dot posta dot it付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 14.50，可以在https://yahoo.com.ph付款。"}, {"role": "assistant", "content": "Yuran是十四令吉五十仙，可以在h t t p s colon slash slash yahoo dot com dot ph付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 541.50，可以在https://example.com/help付款。"}, {"role": "assistant", "content": "Yuran是五百四十一令吉五十仙，可以在h t t p s colon slash slash example dot com slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 500，可以在https://gmail.com付款。"}, {"role": "assistant", "content": "Yuran是五百令吉，可以在h t t p s colon slash slash gmail dot com付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 98.49，可以在www.posta.it/login付款。"}, {"role": "assistant", "content": "Yuran是九十八令吉四十九仙，可以在w w w dot posta dot it slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.4 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 1.90，可以在www.gmail.co.id付款。"}, {"role": "assistant", "content": "Yuran是一令吉九十仙，可以在w w w dot gmail dot co dot id付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 5,000.90，可以在https://yahoo.com.ph付款。"}, {"role": "assistant", "content": "Yuran是五千令吉九十仙，可以在h t t p s colon slash slash yahoo dot com dot ph付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.3°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM26.48，可以在www.sapo.pt/login付款。"}, {"role": "assistant", "content": "Yuran是二十六令吉四十八仙，可以在w w w dot sapo dot pt slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 47.85，可以在www.bank.my/login付款。"}, {"role": "assistant", "content": "Yuran是四十七令吉八十五仙，可以在w w w dot bank dot my slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM41,927.50，可以在https://www.example.com/support付款。"}, {"role": "assistant", "content": "Yuran是四万一千九百二十七令吉五十仙，可以在h t t p s colon slash slash www dot example dot com slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.9°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点九摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.3 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.0 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点零摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.2 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点二摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 80,816.90，可以在https://gmail.com/help付款。"}, {"role": "assistant", "content": "Yuran是八万零八百一十六令吉九十仙，可以在h t t p s colon slash slash gmail dot com slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.8 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 94,893.50，可以在www.mail.fr/login付款。"}, {"role": "assistant", "content": "Yuran是九万四千八百九十三令吉五十仙，可以在w w w dot mail dot fr slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM43，可以在www.example.com付款。"}, {"role": "assistant", "content": "Yuran是四十三令吉，可以在w w w dot example dot com付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 98.64，可以在https://outlook.sa/help付款。"}, {"role": "assistant", "content": "Yuran是九十八令吉六十四仙，可以在h t t p s colon slash slash outlook dot sa slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 76，可以在https://example.com/help付款。"}, {"role": "assistant", "content": "Yuran是七十六令吉，可以在h t t p s colon slash slash example dot com slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.3 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.5 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点五摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.8°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 74.72，可以在https://ziggo.nl/help付款。"}, {"role": "assistant", "content": "Yuran是七十四令吉七十二仙，可以在h t t p s colon slash slash ziggo dot nl slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 39,563，可以在www.posta.it付款。"}, {"role": "assistant", "content": "Yuran是三万九千五百六十三令吉，可以在w w w dot posta dot it付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.8 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM161.90，可以在https://sapo.pt付款。"}, {"role": "assistant", "content": "Yuran是一百六十一令吉九十仙，可以在h t t p s colon slash slash sapo dot pt付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 50，可以在https://gmail.co.id/faq付款。"}, {"role": "assistant", "content": "Yuran是五十令吉，可以在h t t p s colon slash slash gmail dot co dot id slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM68.50，可以在https://www.outlook.sa/support付款。"}, {"role": "assistant", "content": "Yuran是六十八令吉五十仙，可以在h t t p s colon slash slash www dot outlook dot sa slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 23,240，可以在https://gmail.com/help付款。"}, {"role": "assistant", "content": "Yuran是二万三千二百四十令吉，可以在h t t p s colon slash slash gmail dot com slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM55，可以在https://ziggo.nl付款。"}, {"role": "assistant", "content": "Yuran是五十五令吉，可以在h t t p s colon slash slash ziggo dot nl付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM613，可以在https://example.com付款。"}, {"role": "assistant", "content": "Yuran是六百一十三令吉，可以在h t t p s colon slash slash example dot com付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.0°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点零摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM57，可以在https://company.co/faq付款。"}, {"role": "assistant", "content": "Yuran是五十七令吉，可以在h t t p s colon slash slash company dot co slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.6°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点六摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM73,210.50，可以在www.company.co付款。"}, {"role": "assistant", "content": "Yuran是七万三千二百一十令吉五十仙，可以在w w w dot company dot co付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.8 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM71，可以在https://mail.fr付款。"}, {"role": "assistant", "content": "Yuran是七十一令吉，可以在h t t p s colon slash slash mail dot fr付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 113.50，可以在https://yahoo.com.ph付款。"}, {"role": "assistant", "content": "Yuran是一百一十三令吉五十仙，可以在h t t p s colon slash slash yahoo dot com dot ph付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.1 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点一摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.6 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点六摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.8°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.8°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点八摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.5°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点五摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM522.77，可以在https://gmail.com/faq付款。"}, {"role": "assistant", "content": "Yuran是五百二十二令吉七十七仙，可以在h t t p s colon slash slash gmail dot com slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 49.90，可以在https://onet.pl/help付款。"}, {"role": "assistant", "content": "Yuran是四十九令吉九十仙，可以在h t t p s colon slash slash onet dot pl slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.7°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 90,412.50，可以在https://outlook.sa/help付款。"}, {"role": "assistant", "content": "Yuran是九万零四百一十二令吉五十仙，可以在h t t p s colon slash slash outlook dot sa slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.0 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点零摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 44,667.90，可以在https://company.co付款。"}, {"role": "assistant", "content": "Yuran是四万四千六百六十七令吉九十仙，可以在h t t p s colon slash slash company dot co付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM26,904.66，可以在https://mail.fr付款。"}, {"role": "assistant", "content": "Yuran是二万六千九百零四令吉六十六仙，可以在h t t p s colon slash slash mail dot fr付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 705.50，可以在https://outlook.sa付款。"}, {"role": "assistant", "content": "Yuran是七百零五令吉五十仙，可以在h t t p s colon slash slash outlook dot sa付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.9 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点九摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM451.90，可以在https://gmail.co.id/faq付款。"}, {"role": "assistant", "content": "Yuran是四百五十一令吉九十仙，可以在h t t p s colon slash slash gmail dot co dot id slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.3°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.1 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点一摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 481.50，可以在www.scicom.com.my/login付款。"}, {"role": "assistant", "content": "Yuran是四百八十一令吉五十仙，可以在w w w dot scicom dot com dot my slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM178.90，可以在www.ziggo.nl/login付款。"}, {"role": "assistant", "content": "Yuran是一百七十八令吉九十仙，可以在w w w dot ziggo dot nl slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.4 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 79,789.50，可以在https://www.sltnet.lk/support付款。"}, {"role": "assistant", "content": "Yuran是七万九千七百八十九令吉五十仙，可以在h t t p s colon slash slash www dot sltnet dot lk slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.4°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.9 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点九摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM40.14，可以在www.sapo.pt付款。"}, {"role": "assistant", "content": "Yuran是四十令吉十四仙，可以在w w w dot sapo dot pt付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM29,600，可以在https://www.scicom.com.my/support付款。"}, {"role": "assistant", "content": "Yuran是二万九千六百令吉，可以在h t t p s colon slash slash www dot scicom dot com dot my slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM981，可以在www.gmail.com/login付款。"}, {"role": "assistant", "content": "Yuran是九百八十一令吉，可以在w w w dot gmail dot com slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.5 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点五摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.7°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.3 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM24.90，可以在https://www.mail.fr/support付款。"}, {"role": "assistant", "content": "Yuran是二十四令吉九十仙，可以在h t t p s colon slash slash www dot mail dot fr slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 98,589.59，可以在https://www.sltnet.lk/support付款。"}, {"role": "assistant", "content": "Yuran是九万八千五百八十九令吉五十九仙，可以在h t t p s colon slash slash www dot sltnet dot lk slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 69,721.90，可以在https://company.co付款。"}, {"role": "assistant", "content": "Yuran是六万九千七百二十一令吉九十仙，可以在h t t p s colon slash slash company dot co付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.6°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点六摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM51,179，可以在www.company.co/login付款。"}, {"role": "assistant", "content": "Yuran是五万一千一百七十九令吉，可以在w w w dot company dot co slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 986.50，可以在www.web.de付款。"}, {"role": "assistant", "content": "Yuran是九百八十六令吉五十仙，可以在w w w dot web dot de付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM82,539.50，可以在https://posta.it付款。"}, {"role": "assistant", "content": "Yuran是八万二千五百三十九令吉五十仙，可以在h t t p s colon slash slash posta dot it付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM34,925.50，可以在www.posta.it付款。"}, {"role": "assistant", "content": "Yuran是三万四千九百二十五令吉五十仙，可以在w w w dot posta dot it付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.5 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点五摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.4°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM485.50，可以在https://www.yahoo.com/support付款。"}, {"role": "assistant", "content": "Yuran是四百八十五令吉五十仙，可以在h t t p s colon slash slash www dot yahoo dot com slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.2°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点二摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 30,108.50，可以在https://onet.pl付款。"}, {"role": "assistant", "content": "Yuran是三万零一百零八令吉五十仙，可以在h t t p s colon slash slash onet dot pl付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.0°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点零摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.1 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点一摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM47,795.63，可以在https://www.bank.my/support付款。"}, {"role": "assistant", "content": "Yuran是四万七千七百九十五令吉六十三仙，可以在h t t p s colon slash slash www dot bank dot my slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM82，可以在https://mail.fr/help付款。"}, {"role": "assistant", "content": "Yuran是八十二令吉，可以在h t t p s colon slash slash mail dot fr slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.6 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点六摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM23,831，可以在https://gmail.co.id/faq付款。"}, {"role": "assistant", "content": "Yuran是二万三千八百三十一令吉，可以在h t t p s colon slash slash gmail dot co dot id slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.1°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点一摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM613，可以在https://www.company.co/support付款。"}, {"role": "assistant", "content": "Yuran是六百一十三令吉，可以在h t t p s colon slash slash www dot company dot co slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM103.90，可以在https://sapo.pt付款。"}, {"role": "assistant", "content": "Yuran是一百零三令吉九十仙，可以在h t t p s colon slash slash sapo dot pt付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM102.50，可以在https://sltnet.lk/faq付款。"}, {"role": "assistant", "content": "Yuran是一百零二令吉五十仙，可以在h t t p s colon slash slash sltnet dot lk slash faq付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 76，可以在www.posta.it/login付款。"}, {"role": "assistant", "content": "Yuran是七十六令吉，可以在w w w dot posta dot it slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.7 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.9 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点九摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 83，可以在www.bank.my付款。"}, {"role": "assistant", "content": "Yuran是八十三令吉，可以在w w w dot bank dot my付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.7°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是40.0°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是四十点零摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.7 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM64,055，可以在https://www.example.com/support付款。"}, {"role": "assistant", "content": "Yuran是六万四千零五十五令吉，可以在h t t p s colon slash slash www dot example dot com slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.4 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 57,983.90，可以在www.sltnet.lk/login付款。"}, {"role": "assistant", "content": "Yuran是五万七千九百八十三令吉九十仙，可以在w w w dot sltnet dot lk slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 86.90，可以在www.ziggo.nl/login付款。"}, {"role": "assistant", "content": "Yuran是八十六令吉九十仙，可以在w w w dot ziggo dot nl slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 52，可以在https://scicom.com.my/help付款。"}, {"role": "assistant", "content": "Yuran是五十二令吉，可以在h t t p s colon slash slash scicom dot com dot my slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是38.7 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十八点七摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.9°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点九摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 55.50，可以在www.sltnet.lk/login付款。"}, {"role": "assistant", "content": "Yuran是五十五令吉五十仙，可以在w w w dot sltnet dot lk slash login付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是36.6°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十六点六摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM366.88，可以在https://www.gmail.com/support付款。"}, {"role": "assistant", "content": "Yuran是三百六十六令吉八十八仙，可以在h t t p s colon slash slash www dot gmail dot com slash support付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是39.4°C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十九点四摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 42，可以在https://sapo.pt/help付款。"}, {"role": "assistant", "content": "Yuran是四十二令吉，可以在h t t p s colon slash slash sapo dot pt slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Yuran是RM 22，可以在https://example.com/help付款。"}, {"role": "assistant", "content": "Yuran是二十二令吉，可以在h t t p s colon slash slash example dot com slash help付款。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Mandarin Chinese with Malay mixed in. Malaysian Mandarin with Malay mixed in. RM is 令吉, cents 仙.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Mandarin Chinese part is spoken in Mandarin Chinese, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "孩子的suhu是37.3 °C，请先observe。"}, {"role": "assistant", "content": "孩子的suhu是三十七点三摄氏度，请先observe。"}], "lang": "zh-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 22–26 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-two to twenty-six working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM32,075.90க்கு மேல் செலவு செய்தால் 22% cash back கிடைக்கும்."}, {"role": "assistant", "content": "முப்பத்திரண்டாயிரத்து எழுபத்தைந்து ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் இருபத்திரண்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 15 - 24 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order fifteen to twenty-four working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 17–19 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order seventeen to nineteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 18–25 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order eighteen to twenty-five working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 58க்கு மேல் செலவு செய்தால் 10.6% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐம்பத்தெட்டு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் பத்து புள்ளி ஆறு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 8–11 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order eight to eleven working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM25,733க்கு மேல் செலவு செய்தால் 50% cash back கிடைக்கும்."}, {"role": "assistant", "content": "இருபத்தைந்தாயிரத்து எழுநூற்று முப்பத்து மூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் ஐம்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 20–28 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty to twenty-eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 12-19 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twelve to nineteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 693.33க்கு மேல் செலவு செய்தால் 61% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுநூற்று தொண்ணூற்று மூன்று ரிங்கிட் முப்பத்து மூன்று சென்னுக்கு மேல் செலவு செய்தால் அறுபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM58,560.74க்கு மேல் செலவு செய்தால் 63% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐம்பத்தெட்டாயிரத்து ஐந்நூற்று அறுபது ரிங்கிட் எழுபத்து நான்கு சென்னுக்கு மேல் செலவு செய்தால் அறுபத்து மூன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 28-30 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-eight to thirty working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 4-9 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order four to nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 23–31 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-three to thirty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 13 - 18 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order thirteen to eighteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 27-29 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-seven to twenty-nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 24–25 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-four to twenty-five working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 9-16 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order nine to sixteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 21-26 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-one to twenty-six working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 5–10 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order five to ten working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 53,787.48க்கு மேல் செலவு செய்தால் 58.6% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐம்பத்து மூன்றாயிரத்து எழுநூற்று எண்பத்தேழு ரிங்கிட் நாற்பத்தெட்டு சென்னுக்கு மேல் செலவு செய்தால் ஐம்பத்தெட்டு புள்ளி ஆறு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 27-28 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-seven to twenty-eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 72.17க்கு மேல் செலவு செய்தால் 74% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்திரண்டு ரிங்கிட் பதினேழு சென்னுக்கு மேல் செலவு செய்தால் எழுபத்து நான்கு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 5 - 10 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order five to ten working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 956க்கு மேல் செலவு செய்தால் 38% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொள்ளாயிரத்து ஐம்பத்தாறு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் முப்பத்தெட்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 78.70க்கு மேல் செலவு செய்தால் 12.2% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்தெட்டு ரிங்கிட் எழுபது சென்னுக்கு மேல் செலவு செய்தால் பன்னிரண்டு புள்ளி இரண்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 30-33 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order thirty to thirty-three working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 13–21 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order thirteen to twenty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM855.50க்கு மேல் செலவு செய்தால் 30.9% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எண்ணூற்று ஐம்பத்தைந்து ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் முப்பது புள்ளி ஒன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 27-31 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-seven to thirty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM14,635க்கு மேல் செலவு செய்தால் 21% cash back கிடைக்கும்."}, {"role": "assistant", "content": "பதினான்காயிரத்து அறுநூற்று முப்பத்தைந்து ரிங்கிட்டுக்கு மேல் செலவு செய்தால் இருபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 762.15க்கு மேல் செலவு செய்தால் 55% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுநூற்று அறுபத்திரண்டு ரிங்கிட் பதினைந்து சென்னுக்கு மேல் செலவு செய்தால் ஐம்பத்தைந்து சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM54,680.90க்கு மேல் செலவு செய்தால் 49% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐம்பத்து நான்காயிரத்து அறுநூற்று எண்பது ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் நாற்பத்தொன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM713க்கு மேல் செலவு செய்தால் 32.7% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுநூற்று பதிமூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் முப்பத்திரண்டு புள்ளி ஏழு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 30-31 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order thirty to thirty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 3 - 5 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order three to five working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 523.90க்கு மேல் செலவு செய்தால் 31% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐந்நூற்று இருபத்து மூன்று ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் முப்பத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM942.50க்கு மேல் செலவு செய்தால் 64% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொள்ளாயிரத்து நாற்பத்திரண்டு ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் அறுபத்து நான்கு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM29.90க்கு மேல் செலவு செய்தால் 4.8% cash back கிடைக்கும்."}, {"role": "assistant", "content": "இருபத்தொன்பது ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் நான்கு புள்ளி எட்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 98,203க்கு மேல் செலவு செய்தால் 87.9% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொண்ணூற்று எட்டாயிரத்து இருநூற்று மூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் எண்பத்தேழு புள்ளி ஒன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM69.19க்கு மேல் செலவு செய்தால் 9% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுபத்தொன்பது ரிங்கிட் பத்தொன்பது சென்னுக்கு மேல் செலவு செய்தால் ஒன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 89,105க்கு மேல் செலவு செய்தால் 39% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எண்பத்தொன்பதாயிரத்து நூற்று ஐந்து ரிங்கிட்டுக்கு மேல் செலவு செய்தால் முப்பத்தொன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 12–17 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twelve to seventeen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 7-17 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order seven to seventeen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM929.90க்கு மேல் செலவு செய்தால் 69.1% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொள்ளாயிரத்து இருபத்தொன்பது ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் அறுபத்தொன்பது புள்ளி ஒன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 18-28 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order eighteen to twenty-eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 6 - 16 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order six to sixteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM29,498க்கு மேல் செலவு செய்தால் 80% cash back கிடைக்கும்."}, {"role": "assistant", "content": "இருபத்தொன்பதாயிரத்து நானூற்று தொண்ணூற்று எட்டு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் எண்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 18-21 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order eighteen to twenty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM67.76க்கு மேல் செலவு செய்தால் 97% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுபத்தேழு ரிங்கிட் எழுபத்தாறு சென்னுக்கு மேல் செலவு செய்தால் தொண்ணூற்று ஏழு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 16 - 19 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order sixteen to nineteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 21-23 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-one to twenty-three working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 79,456.37க்கு மேல் செலவு செய்தால் 36% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்தொன்பதாயிரத்து நானூற்று ஐம்பத்தாறு ரிங்கிட் முப்பத்தேழு சென்னுக்கு மேல் செலவு செய்தால் முப்பத்தாறு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM13க்கு மேல் செலவு செய்தால் 69% cash back கிடைக்கும்."}, {"role": "assistant", "content": "பதிமூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் அறுபத்தொன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM559.50க்கு மேல் செலவு செய்தால் 45% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐந்நூற்று ஐம்பத்தொன்பது ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் நாற்பத்தைந்து சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 75,329க்கு மேல் செலவு செய்தால் 77% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்தைந்தாயிரத்து முந்நூற்று இருபத்தொன்பது ரிங்கிட்டுக்கு மேல் செலவு செய்தால் எழுபத்தேழு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 78.58க்கு மேல் செலவு செய்தால் 60% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்தெட்டு ரிங்கிட் ஐம்பத்தெட்டு சென்னுக்கு மேல் செலவு செய்தால் அறுபது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 6 - 9 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order six to nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 22–28 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-two to twenty-eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM25.90க்கு மேல் செலவு செய்தால் 88.2% cash back கிடைக்கும்."}, {"role": "assistant", "content": "இருபத்தைந்து ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் எண்பத்தெட்டு புள்ளி இரண்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 140.35க்கு மேல் செலவு செய்தால் 5% cash back கிடைக்கும்."}, {"role": "assistant", "content": "நூற்று நாற்பது ரிங்கிட் முப்பத்தைந்து சென்னுக்கு மேல் செலவு செய்தால் ஐந்து சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 931.90க்கு மேல் செலவு செய்தால் 71% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொள்ளாயிரத்து முப்பத்தொன்று ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் எழுபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 3 - 9 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order three to nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM28.90க்கு மேல் செலவு செய்தால் 99% cash back கிடைக்கும்."}, {"role": "assistant", "content": "இருபத்தெட்டு ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் தொண்ணூற்று ஒன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 468க்கு மேல் செலவு செய்தால் 30% cash back கிடைக்கும்."}, {"role": "assistant", "content": "நானூற்று அறுபத்தெட்டு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் முப்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 19 - 28 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order nineteen to twenty-eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 27–33 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-seven to thirty-three working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 28–29 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-eight to twenty-nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 4-14 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order four to fourteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 2-6 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order two to six working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 8 - 14 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order eight to fourteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 5-8 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order five to eight working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 20 - 21 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty to twenty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM43.90க்கு மேல் செலவு செய்தால் 20% cash back கிடைக்கும்."}, {"role": "assistant", "content": "நாற்பத்து மூன்று ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் இருபது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 14 - 16 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order fourteen to sixteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM12,584க்கு மேல் செலவு செய்தால் 64% cash back கிடைக்கும்."}, {"role": "assistant", "content": "பன்னிரண்டாயிரத்து ஐந்நூற்று எண்பத்து நான்கு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் அறுபத்து நான்கு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 653க்கு மேல் செலவு செய்தால் 4.1% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுநூற்று ஐம்பத்து மூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் நான்கு புள்ளி ஒன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 3-6 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order three to six working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 516க்கு மேல் செலவு செய்தால் 82% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐந்நூற்று பதினாறு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் எண்பத்திரண்டு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 26-31 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-six to thirty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 15 - 16 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order fifteen to sixteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 5க்கு மேல் செலவு செய்தால் 55% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐந்து ரிங்கிட்டுக்கு மேல் செலவு செய்தால் ஐம்பத்தைந்து சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM97,293.50க்கு மேல் செலவு செய்தால் 84% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொண்ணூற்று ஏழாயிரத்து இருநூற்று தொண்ணூற்று மூன்று ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் எண்பத்து நான்கு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM41.50க்கு மேல் செலவு செய்தால் 36.4% cash back கிடைக்கும்."}, {"role": "assistant", "content": "நாற்பத்தொன்று ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் முப்பத்தாறு புள்ளி நான்கு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 66,713க்கு மேல் செலவு செய்தால் 71% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுபத்தாறாயிரத்து எழுநூற்று பதிமூன்று ரிங்கிட்டுக்கு மேல் செலவு செய்தால் எழுபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 5-7 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order five to seven working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM77,875க்கு மேல் செலவு செய்தால் 21% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எழுபத்தேழாயிரத்து எண்ணூற்று எழுபத்தைந்து ரிங்கிட்டுக்கு மேல் செலவு செய்தால் இருபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 9-18 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order nine to eighteen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM11,835.90க்கு மேல் செலவு செய்தால் 94.7% cash back கிடைக்கும்."}, {"role": "assistant", "content": "பதினொன்றாயிரத்து எண்ணூற்று முப்பத்தைந்து ரிங்கிட் தொண்ணூறு சென்னுக்கு மேல் செலவு செய்தால் தொண்ணூற்று நான்கு புள்ளி ஏழு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM46க்கு மேல் செலவு செய்தால் 97% cash back கிடைக்கும்."}, {"role": "assistant", "content": "நாற்பத்தாறு ரிங்கிட்டுக்கு மேல் செலவு செய்தால் தொண்ணூற்று ஏழு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 23 - 31 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order twenty-three to thirty-one working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 7–10 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order seven to ten working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 52,407.50க்கு மேல் செலவு செய்தால் 20% cash back கிடைக்கும்."}, {"role": "assistant", "content": "ஐம்பத்திரண்டாயிரத்து நானூற்று ஏழு ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் இருபது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 97,930க்கு மேல் செலவு செய்தால் 96% cash back கிடைக்கும்."}, {"role": "assistant", "content": "தொண்ணூற்று ஏழாயிரத்து தொள்ளாயிரத்து முப்பது ரிங்கிட்டுக்கு மேல் செலவு செய்தால் தொண்ணூற்று ஆறு சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 30–32 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order thirty to thirty-two working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 15–17 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order fifteen to seventeen working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM894.50க்கு மேல் செலவு செய்தால் 61% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எண்ணூற்று தொண்ணூற்று நான்கு ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் அறுபத்தொன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் order 3–9 working days-ல் வந்துவிடும்."}, {"role": "assistant", "content": "உங்கள் order three to nine working days-ல் வந்துவிடும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM 89,945க்கு மேல் செலவு செய்தால் 43% cash back கிடைக்கும்."}, {"role": "assistant", "content": "எண்பத்தொன்பதாயிரத்து தொள்ளாயிரத்து நாற்பத்தைந்து ரிங்கிட்டுக்கு மேல் செலவு செய்தால் நாற்பத்து மூன்று சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with English mixed in. Malaysian Tamil with English mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the English part in English. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM663.50க்கு மேல் செலவு செய்தால் 53.9% cash back கிடைக்கும்."}, {"role": "assistant", "content": "அறுநூற்று அறுபத்து மூன்று ரிங்கிட் ஐம்பது சென்னுக்கு மேல் செலவு செய்தால் ஐம்பத்து மூன்று புள்ளி ஒன்பது சதவீதம் cash back கிடைக்கும்."}], "lang": "ta-en", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:05 மணி முதல் 7.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது பூஜ்ஜியம் ஐந்து மணி முதல் ஏழு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 5.45 மணி முதல் 14:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஐந்து நாற்பத்தைந்து மணி முதல் பதினான்கு மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 89%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எண்பத்தொன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 84%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எண்பத்து நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 16.15 மணி முதல் 2:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினாறு பதினைந்து மணி முதல் இரண்டு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 16:15 மணி முதல் 19:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினாறு பதினைந்து மணி முதல் பத்தொன்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 11:30 மணி முதல் 15:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினொன்று முப்பது மணி முதல் பதினைந்து முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 80%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எண்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 20:05 மணி முதல் 8:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபது பூஜ்ஜியம் ஐந்து மணி முதல் எட்டு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 9%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஒன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 18.45 மணி முதல் 18:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினெட்டு நாற்பத்தைந்து மணி முதல் பதினெட்டு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 3.45 மணி முதல் 10.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat மூன்று நாற்பத்தைந்து மணி முதல் பத்து நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 42.4%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்திரண்டு புள்ளி நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 14:45 மணி முதல் 8:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினான்கு நாற்பத்தைந்து மணி முதல் எட்டு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 11:30 மணி முதல் 6.30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினொன்று முப்பது மணி முதல் ஆறு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 21:05 மணி முதல் 17:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்தொன்று பூஜ்ஜியம் ஐந்து மணி முதல் பதினேழு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 46%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்தாறு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 3:00 மணி முதல் 13:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat மூன்று மணி முதல் பதிமூன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 31.4%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் முப்பத்தொன்று புள்ளி நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 7:45 மணி முதல் 22.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஏழு நாற்பத்தைந்து மணி முதல் இருபத்திரண்டு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 18:00 மணி முதல் 20:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினெட்டு மணி முதல் இருபது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 22.45 மணி முதல் 7.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்திரண்டு நாற்பத்தைந்து மணி முதல் ஏழு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 5.05 மணி முதல் 22:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஐந்து பூஜ்ஜியம் ஐந்து மணி முதல் இருபத்திரண்டு மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12.05 மணி முதல் 21:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு பூஜ்ஜியம் ஐந்து மணி முதல் இருபத்தொன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19:45 மணி முதல் 7:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது நாற்பத்தைந்து மணி முதல் ஏழு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12:15 மணி முதல் 12.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு பதினைந்து மணி முதல் பன்னிரண்டு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 17:45 மணி முதல் 4:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினேழு நாற்பத்தைந்து மணி முதல் நான்கு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 15:00 மணி முதல் 8.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினைந்து மணி முதல் எட்டு பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 18:15 மணி முதல் 21:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினெட்டு பதினைந்து மணி முதல் இருபத்தொன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:15 மணி முதல் 11.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது பதினைந்து மணி முதல் பதினொன்று பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 16:30 மணி முதல் 12:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினாறு முப்பது மணி முதல் பன்னிரண்டு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 3.30 மணி முதல் 23:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat மூன்று முப்பது மணி முதல் இருபத்து மூன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:00 மணி முதல் 17:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது மணி முதல் பதினேழு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 98%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் தொண்ணூற்று எட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 99%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் தொண்ணூற்று ஒன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 54%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஐம்பத்து நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 70.5%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபது புள்ளி ஐந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 40%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 22.15 மணி முதல் 13:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்திரண்டு பதினைந்து மணி முதல் பதிமூன்று மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12:05 மணி முதல் 12:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு பூஜ்ஜியம் ஐந்து மணி முதல் பன்னிரண்டு மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 26.4%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்தாறு புள்ளி நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 22%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்திரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 19%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பத்தொன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 77%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்தேழு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 22.45 மணி முதல் 11.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்திரண்டு நாற்பத்தைந்து மணி முதல் பதினொன்று நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 22:05 மணி முதல் 12.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்திரண்டு பூஜ்ஜியம் ஐந்து மணி முதல் பன்னிரண்டு பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 17:15 மணி முதல் 18:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினேழு பதினைந்து மணி முதல் பதினெட்டு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19:30 மணி முதல் 7:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது முப்பது மணி முதல் ஏழு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 4.05 மணி முதல் 13.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat நான்கு பூஜ்ஜியம் ஐந்து மணி முதல் பதிமூன்று நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 4.4%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நான்கு புள்ளி நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12.30 மணி முதல் 23:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு முப்பது மணி முதல் இருபத்து மூன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 7:00 மணி முதல் 20:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஏழு மணி முதல் இருபது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12:30 மணி முதல் 19:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு முப்பது மணி முதல் பத்தொன்பது முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 48%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்தெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 49.5%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்தொன்பது புள்ளி ஐந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 5%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஐந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 76.1%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்தாறு புள்ளி ஒன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 21:30 மணி முதல் 13.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்தொன்று முப்பது மணி முதல் பதிமூன்று பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 17:05 மணி முதல் 11.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினேழு பூஜ்ஜியம் ஐந்து மணி முதல் பதினொன்று பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 74.6%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்து நான்கு புள்ளி ஆறு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 1:05 மணி முதல் 6.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்று பூஜ்ஜியம் ஐந்து மணி முதல் ஆறு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 4.45 மணி முதல் 2:15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat நான்கு நாற்பத்தைந்து மணி முதல் இரண்டு பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 1.30 மணி முதல் 23.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்று முப்பது மணி முதல் இருபத்து மூன்று பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 28%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்தெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 68.4%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் அறுபத்தெட்டு புள்ளி நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 15:00 மணி முதல் 9:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினைந்து மணி முதல் ஒன்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 23:45 மணி முதல் 23.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்து மூன்று நாற்பத்தைந்து மணி முதல் இருபத்து மூன்று பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 3.7%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் மூன்று புள்ளி ஏழு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 1.45 மணி முதல் 16:15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்று நாற்பத்தைந்து மணி முதல் பதினாறு பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 97.2%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் தொண்ணூற்று ஏழு புள்ளி இரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 83%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எண்பத்து மூன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 32%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் முப்பத்திரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 45%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்தைந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 1.1%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஒரு புள்ளி ஒன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 58.7%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஐம்பத்தெட்டு புள்ளி ஏழு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 11.15 மணி முதல் 23:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினொன்று பதினைந்து மணி முதல் இருபத்து மூன்று நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 38%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் முப்பத்தெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 68%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் அறுபத்தெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 6.15 மணி முதல் 12.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஆறு பதினைந்து மணி முதல் பன்னிரண்டு பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 11.6%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பதினொன்று புள்ளி ஆறு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 14:00 மணி முதல் 16:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினான்கு மணி முதல் பதினாறு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:15 மணி முதல் 21.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது பதினைந்து மணி முதல் இருபத்தொன்று நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19.15 மணி முதல் 3:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது பதினைந்து மணி முதல் மூன்று நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 20.9%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபது புள்ளி ஒன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19.45 மணி முதல் 15.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது நாற்பத்தைந்து மணி முதல் பதினைந்து பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:05 மணி முதல் 5.15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது பூஜ்ஜியம் ஐந்து மணி முதல் ஐந்து பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 18%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பதினெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 23:05 மணி முதல் 14.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்து மூன்று பூஜ்ஜியம் ஐந்து மணி முதல் பதினான்கு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 18:45 மணி முதல் 2.45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினெட்டு நாற்பத்தைந்து மணி முதல் இரண்டு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 12.2%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பன்னிரண்டு புள்ளி இரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 62%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் அறுபத்திரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 15:05 மணி முதல் 23:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதினைந்து பூஜ்ஜியம் ஐந்து மணி முதல் இருபத்து மூன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 74%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்து நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 78%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்தெட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 23%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்து மூன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 14%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பதினான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 39%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் முப்பத்தொன்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 4:00 மணி முதல் 13.30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat நான்கு மணி முதல் பதிமூன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 43%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்து மூன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 5:30 மணி முதல் 19.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஐந்து முப்பது மணி முதல் பத்தொன்பது பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 13:15 மணி முதல் 22:05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பதிமூன்று பதினைந்து மணி முதல் இருபத்திரண்டு பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 66%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் அறுபத்தாறு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 47.8%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் நாற்பத்தேழு புள்ளி எட்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 12.05 மணி முதல் 10:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பன்னிரண்டு பூஜ்ஜியம் ஐந்து மணி முதல் பத்து நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 11.5%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் பதினொன்று புள்ளி ஐந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 25%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்தைந்து சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 4:15 மணி முதல் 2.30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat நான்கு பதினைந்து மணி முதல் இரண்டு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 21.30 மணி முதல் 13:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்தொன்று முப்பது மணி முதல் பதிமூன்று மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 52.7%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஐம்பத்திரண்டு புள்ளி ஏழு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 23.45 மணி முதல் 6:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்து மூன்று நாற்பத்தைந்து மணி முதல் ஆறு மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 64%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் அறுபத்து நான்கு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9.45 மணி முதல் 16.30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது நாற்பத்தைந்து மணி முதல் பதினாறு முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 74.2%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் எழுபத்து நான்கு புள்ளி இரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 23.6%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்து மூன்று புள்ளி ஆறு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 23.15 மணி முதல் 2:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபத்து மூன்று பதினைந்து மணி முதல் இரண்டு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19:15 மணி முதல் 10:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது பதினைந்து மணி முதல் பத்து முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 21%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் இருபத்தொன்று சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 9:00 மணி முதல் 15:15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஒன்பது மணி முதல் பதினைந்து பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 20:30 மணி முதல் 6:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat இருபது முப்பது மணி முதல் ஆறு நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 92%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் தொண்ணூற்று இரண்டு சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19:45 மணி முதல் 11:15 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது நாற்பத்தைந்து மணி முதல் பதினொன்று பதினைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Denda outstanding-ன் 50%, tolong bayar tepat pada masa."}, {"role": "assistant", "content": "Denda outstanding-ன் ஐம்பது சதவீதம், tolong bayar tepat pada masa."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19.05 மணி முதல் 11:30 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது பூஜ்ஜியம் ஐந்து மணி முதல் பதினொன்று முப்பது மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 6.45 மணி முதல் 3.05 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஆறு நாற்பத்தைந்து மணி முதல் மூன்று பூஜ்ஜியம் ஐந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 7.45 மணி முதல் 6:00 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat ஏழு நாற்பத்தைந்து மணி முதல் ஆறு மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. The text is code-switched: mostly Tamil with Malay mixed in. Malaysian Tamil with Malay mixed in. RM is ரிங்கிட், cents சென்.\nRules:\n- Expand everything that is not speakable as written: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP numbers digit by digit, dates, times, percentages, decimals, ordinals, units, email addresses and URLs.\n- Read each number in the language of the words AROUND it: a number inside the Tamil part is spoken in Tamil, a number inside the Malay part in Malay. Do not switch the sentence to one language.\n- Never answer questions, never add, drop, reorder or translate words. Every word that is already speakable stays exactly as written, in the language it was written in, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pejabat 19:05 மணி முதல் 19:45 மணி வரை buka."}, {"role": "assistant", "content": "Pejabat பத்தொன்பது பூஜ்ஜியம் ஐந்து மணி முதல் பத்தொன்பது நாற்பத்தைந்து மணி வரை buka."}], "lang": "ta-ms", "source": "template"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "My bill is RM125.40 this month, why so expensive?"}, {"role": "assistant", "content": "My bill is one hundred twenty-five ringgit forty sen this month, why so expensive?"}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please transfer RM1,200.00 to the clinic for the procedure."}, {"role": "assistant", "content": "Please transfer one thousand two hundred ringgit to the clinic for the procedure."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The medical insurance payout was RM85,000.00."}, {"role": "assistant", "content": "The medical insurance payout was eighty-five thousand ringgit."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Your account has been debited RM150.00 for the monthly insurance premium."}, {"role": "assistant", "content": "Your account has been debited ringgit one hundred fifty for the monthly insurance premium."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "I'm trying to top up my credit but the order number 5 0 6 1 2 is showing as invalid."}, {"role": "assistant", "content": "I'm trying to top up my credit but the order number five zero six one two is showing as invalid."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Doctor is only available for consultation starting 2.30pm."}, {"role": "assistant", "content": "Doctor is only available for consultation starting two thirty p m."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Your data plan is almost finished, top up now to get 10-20GB extra data."}, {"role": "assistant", "content": "Your data plan is almost finished, top up now to get ten to twenty gigabytes extra data."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please ensure your account has at least RM500 for the monthly maintenance fee."}, {"role": "assistant", "content": "Please ensure your account has at least five hundred ringgit for the monthly maintenance fee."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Please download your medical report from the patient portal: https://myhealth.gov.my/reports/12345"}, {"role": "assistant", "content": "Please download your medical report from the patient portal: h t t p s colon slash slash my health dot gov dot my slash reports slash one two three four five"}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Your telco rebate of RM5.50 sen has been credited, check it out at https://celcom.com.my/my-account"}, {"role": "assistant", "content": "Your telco rebate of five ringgit fifty sen has been credited, check it out at h t t p s colon slash slash celcom dot com dot m y slash my hyphen account"}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Call 03-21668888 to reschedule your 2pm appointment for 18/11."}, {"role": "assistant", "content": "Call zero three-two one six six eight eight eight eight to reschedule your two pm appointment for eighteen slash eleven."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in English, ready to be synthesized. Malaysian English: RM is ringgit, sen for cents.\nRules:\n- Expand everything that is not speakable as written into English words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in English even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "The stock price dropped by -1.25% after the news broke."}, {"role": "assistant", "content": "The stock price dropped by minus one point two five percent after the news broke."}], "lang": "en", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Liputan rangkaian telekomunikasi di kawasan pedalaman kini semakin stabil dan luas."}, {"role": "assistant", "content": "Liputan rangkaian telekomunikasi di kawasan pedalaman kini semakin stabil dan luas."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Klinik ni buka dari pukul 9 pagi sampai 5 petang setiap hari."}, {"role": "assistant", "content": "Klinik ni buka dari pukul sembilan pagi sampai lima petang setiap hari."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kos pembedahan jantung itu dianggarkan mencecah RM45,000 mengikut sebut harga hospital."}, {"role": "assistant", "content": "Kos pembedahan jantung itu dianggarkan mencecah empat puluh lima ribu ringgit mengikut sebut harga hospital."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Dana kecemasan syarikat telah meningkat kepada RM8,750,000 bagi menghadapi krisis ekonomi."}, {"role": "assistant", "content": "Dana kecemasan syarikat telah meningkat kepada lapan juta tujuh ratus lima puluh ribu ringgit bagi menghadapi krisis ekonomi."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Jumlah pemindahan ke akaun Maybank adalah RM1,250.75."}, {"role": "assistant", "content": "Jumlah pemindahan ke akaun Maybank adalah satu ribu dua ratus lima puluh ringgit tujuh puluh lima sen."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "RM50.00 telah dipindahkan ke akaun 123456789012 via DuitNow."}, {"role": "assistant", "content": "lima puluh ringgit sifar sen telah dipindahkan ke akaun satu dua tiga empat lima enam tujuh lapan sembilan sifar satu dua via DuitNow."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Boleh saya tahu no akaun bank encik? Saya nak check status transfer RM1,200.50 tu, 1 0 2 9 3 8 4 7 5 6."}, {"role": "assistant", "content": "Boleh saya tahu no akaun bank encik? Saya nak check status transfer ringgit Malaysia satu ribu dua ratus ringgit lima puluh sen tu, satu kosong dua sembilan tiga lapan empat tujuh lima enam."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ubat ni kena ambil ikut preskripsi, no order 5 5 0 1 2 3."}, {"role": "assistant", "content": "Ubat ni kena ambil ikut preskripsi, no order lima lima sifar satu dua tiga."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sila sahkan no pesanan 9 0 4 3 2 1 untuk temu janji doktor esok."}, {"role": "assistant", "content": "Sila sahkan no pesanan sembilan sifar empat tiga dua satu untuk temu janji doktor esok."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Janji temu doktor anda adalah pada 15 Okt jam 0930, sila bawa kad pengenalan."}, {"role": "assistant", "content": "Janji temu doktor anda adalah pada lima belas Oktober jam kosong sembilan tiga puluh, sila bawa kad pengenalan."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Tunggu kejap, saya check balance bank pukul 2.30 petang nanti."}, {"role": "assistant", "content": "Tunggu kejap, saya check balance bank pukul dua tiga puluh petang nanti."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Jumlah pembiayaan perumahan telah meningkat sebanyak RM500k sejak dekad 2010-an."}, {"role": "assistant", "content": "Jumlah pembiayaan perumahan telah meningkat sebanyak lima ratus ribu ringgit sejak dekad dua ribu sepuluh-an."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kredit RM10.50 akan dikreditkan semula ke akaun anda sebelum hujung 2024."}, {"role": "assistant", "content": "Kredit sepuluh ringgit lima puluh sen akan dikreditkan semula ke akaun anda sebelum hujung dua ribu dua puluh empat."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Sila hantar laporan perubatan ke dr.azman@kkm.gov.my sebelum jam 5 petang esok."}, {"role": "assistant", "content": "Sila hantar laporan perubatan ke dr dot azman at kkm dot gov dot my sebelum jam lima petang esok."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Malay, ready to be synthesized. Bahasa Melayu Malaysia (RM = ringgit, sen).\nRules:\n- Expand everything that is not speakable as written into Malay words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Malay even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kod pengesahan OTP anda ialah 882931, sah sehingga jam 11:45am hari ini."}, {"role": "assistant", "content": "Kod pengesahan OTP anda ialah lapan lapan dua sembilan tiga satu, sah sehingga jam sebelas empat puluh lima pagi hari ini."}], "lang": "ms", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Saldo tabungan saya tinggal Rp 1.250.000 setelah bayar kontrakan."}, {"role": "assistant", "content": "Saldo tabungan saya tinggal satu juta dua ratus lima puluh ribu rupiah setelah bayar kontrakan."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pembayaran obat-obatan di apotek totalnya jadi Rp1.120.700."}, {"role": "assistant", "content": "Pembayaran obat-obatan di apotek totalnya jadi satu juta seratus dua puluh ribu tujuh ratus rupiah."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mohon konfirmasi nomor rekening 122009876543 untuk pengiriman dana sebesar Rp5.000.000."}, {"role": "assistant", "content": "Mohon konfirmasi nomor rekening satu dua dua nol nol sembilan delapan tujuh enam lima empat tiga untuk pengiriman dana sebesar lima juta rupiah."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Password sementara untuk akses portal kesehatan Anda adalah 903821."}, {"role": "assistant", "content": "Password sementara untuk akses portal kesehatan Anda adalah sembilan nol tiga delapan dua satu."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "CS Telkomsel bisa dihubungi via chat mulai jam 9 pagi sampai 5 sore."}, {"role": "assistant", "content": "CS Telkomsel bisa dihubungi via chat mulai jam sembilan pagi sampai lima sore."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Paket data 50GB sekarang lagi promo cuma 100rb buat pengguna baru."}, {"role": "assistant", "content": "Paket data lima puluh gigabita sekarang lagi promo cuma seratus ribu rupiah buat pengguna baru."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Tolong transfer Rp2.500.000 ke rekening BCA saya ya."}, {"role": "assistant", "content": "Tolong transfer dua juta lima ratus ribu rupiah ke rekening BCA saya ya."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Rekam medis pasien dari tahun 1990-an sudah didigitalisasi semua."}, {"role": "assistant", "content": "Rekam medis pasien dari tahun seribu sembilan ratus sembilan puluhan sudah didigitalisasi semua."}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Indonesian, ready to be synthesized. Bahasa Indonesia (Rp = rupiah).\nRules:\n- Expand everything that is not speakable as written into Indonesian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Indonesian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Hasil lab sudah keluar, silakan download di https://prodia.co.id/portal-pasien"}, {"role": "assistant", "content": "Hasil lab sudah keluar, silakan download di prodia titik co titik id garis miring portal strip pasien"}], "lang": "id", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "医生建议我每天服用 2 次药物，每次 1 粒。"}, {"role": "assistant", "content": "医生建议我每天服用 二 次药物，每次 一 粒。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "这份医疗报告的打印费用是 RM5.00。"}, {"role": "assistant", "content": "这份医疗报告的打印费用是 五令吉零分。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的医疗记录编号是 HN-7 7 3 1 0，请在柜台出示。"}, {"role": "assistant", "content": "您的医疗记录编号是 HN-七 七 三 一 零，请在柜台出示。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的快递单号是 MY-5 5 4 0 1 2 8，预计明天送到。"}, {"role": "assistant", "content": "您的快递单号是 M Y 杠 五 五 四 零 一 二 八，预计明天送到。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的银行账户将在2024-05-12之前扣除RM50的年费。"}, {"role": "assistant", "content": "您的银行账户将在二零二四年五月十二日之前扣除五十令吉的年费。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您的信用卡在 23:00 之后刷卡失败，请联系客服。"}, {"role": "assistant", "content": "如果您的信用卡在 二十三点零零 之后刷卡失败，请联系客服。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您在 1st day 激活 SIM 卡，可以获得额外流量。"}, {"role": "assistant", "content": "如果您在 第一天 激活 SIM 卡，可以获得额外流量。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "请在 1st floor 的药房领取您的药物。"}, {"role": "assistant", "content": "请在 第一层 floor 的药房领取您的药物。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "如果您在 24 hr 内没有确认交易，该申请将自动取消。"}, {"role": "assistant", "content": "如果您在 二十四小时 内没有确认交易，该申请将自动取消。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "这个 5G 预付配套每月只需 RM35，还送 1 TB 云端存储。"}, {"role": "assistant", "content": "这个 五G 预付配套每月只需 三十五令吉，还送 一太字节 云端存储。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的贷款申请已通过，年利率为 4.25%，首期款 RM1,200 请于明天缴付。"}, {"role": "assistant", "content": "您的贷款申请已通过，年利率为 百分之四点二五，首期款 一千二百令吉 请于明天缴付。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户被锁定，请输入代码 9910 并拨打 019-7778889 解锁。"}, {"role": "assistant", "content": "您的账户被锁定，请输入代码 九九一零 并拨打 零一九-七七七八八八九 解锁。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Mandarin Chinese, ready to be synthesized. Simplified Chinese as used in Malaysia (RM = 令吉).\nRules:\n- Expand everything that is not speakable as written into Mandarin Chinese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Mandarin Chinese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "您的账户余额不足，导致扣款失败，目前余额为-RM15.50。"}, {"role": "assistant", "content": "您的账户余额不足，导致扣款失败，目前余额为负十五点五零令吉。"}], "lang": "zh", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இணையதள சேவை இன்று மாலை வரை தடை செய்யப்பட்டிருக்கும்"}, {"role": "assistant", "content": "இணையதள சேவை இன்று மாலை வரை தடை செய்யப்பட்டிருக்கும்"}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் கார்டு பிளாக் செய்யப்பட்டுள்ளது, உடனடியாக +6016-2233445 என்ற எண்ணுக்கு அழைக்கவும்."}, {"role": "assistant", "content": "உங்கள் கார்டு பிளாக் செய்யப்பட்டுள்ளது, உடனடியாக பிளஸ் ஆறு பூஜ்ஜியம் ஒன்று ஆறு ஹைபன் இரண்டு இரண்டு மூன்று மூன்று நான்கு நான்கு ஐந்து என்ற எண்ணுக்கு அழைக்கவும்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் டேட்டா பேக் 20-11-2023 அன்று முடிவடையும், தயவுசெய்து ரீலோட் செய்யவும்."}, {"role": "assistant", "content": "உங்கள் டேட்டா பேக் இருபது நவம்பர் இரண்டாயிரத்து இருபத்தி மூன்று அன்று முடிவடையும், தயவுசெய்து ரீலோட் செய்யவும்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் வங்கி கணக்கு 10-05-2023 அன்று வெற்றிகரமாக செயல்படுத்தப்பட்டது."}, {"role": "assistant", "content": "உங்கள் வங்கி கணக்கு பத்து மே இரு हज़ार இருபத்தி மூன்று அன்று வெற்றிகரமாக செயல்படுத்தப்பட்டது."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "மருந்து கடை இன்று மாலை 6:00pm வரை மட்டுமே திறந்திருக்கும்."}, {"role": "assistant", "content": "மருந்து கடை இன்று மாலை ஆறு மணி வரை மட்டுமே திறந்திருக்கும்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களது 5th payment நிலுவையில் உள்ளது, விரைவாக RM100 செலுத்துங்கள்."}, {"role": "assistant", "content": "உங்களது ஐந்தாவது payment நிலுவையில் உள்ளது, விரைவாக நூறு ரிங்கிட் செலுத்துங்கள்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "கடந்த 2 தசாப்தங்களாக வங்கி சேவைகள் டிஜிட்டலாக மாறிவிட்டன."}, {"role": "assistant", "content": "கடந்த இரண்டு தசாப்தங்களாக வங்கி சேவைகள் டிஜிட்டலாக மாறிவிட்டன."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்கள் லோன் அப்ளிகேஷன் பற்றிய விவரங்களுக்கு support@maybank.com.my என்ற மின்னஞ்சலுக்குத் தொடர்பு கொள்ளவும்."}, {"role": "assistant", "content": "உங்கள் லோன் அப்ளிகேஷன் பற்றிய விவரங்களுக்கு support at maybank dot com dot my என்ற மின்னஞ்சலுக்குத் தொடர்பு கொள்ளவும்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய அப்பாயிண்ட்மென்ட்டை https://www.mysejahtera.com.my மூலமாக கன்ஃபர்ம் செய்யவும்."}, {"role": "assistant", "content": "உங்களுடைய அப்பாயிண்ட்மென்ட்டை ஹெச்டிடிபிஎஸ் கோலன் ஸ்லாஷ் ஸ்லாஷ் www டாட் mysejahtera டாட் com டாட் my மூலமாக கன்ஃபர்ம் செய்யவும்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "மருத்துவமனை பில் RM1,200, இதை 15/11/2023 தேதிக்குள்ள பே பண்ணுங்க."}, {"role": "assistant", "content": "மருத்துவமனை பில் ஒரு ஆயிரத்து இருநூறு ரிங்கிட், இதை பதினைந்து நவம்பர் இரண்டாயிரத்து இருபத்தி மூன்று தேதிக்குள்ள பே பண்ணுங்க."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil, ready to be synthesized. Tamil as used in Malaysia (RM = ரிங்கிட்).\nRules:\n- Expand everything that is not speakable as written into Tamil words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "டாக்டர் 3:15pm-க்கு ஃப்ரீயா இருப்பாங்க, உங்களுக்கு 017-9876543 நம்பர்ல கால் பண்றோம்."}, {"role": "assistant", "content": "டாக்டர் மூன்று பதினைந்து பி.எம்-க்கு ஃப்ரீயா இருப்பாங்க, உங்களுக்கு பூஜ்யம் ஒன்று ஏழு ஸ்ட்ரைக் ஒன்பது எட்டு ஏழு ஆறு ஐந்து நான்கு மூன்று நம்பர்ல கால் பண்றோம்."}], "lang": "ta", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இன்று கிளினிக்கில் 40 நோயாளிகள் பதிவு செய்துள்ளனர்."}, {"role": "assistant", "content": "இன்று கிளினிக்கில் நாற்பது நோயாளிகள் பதிவு செய்துள்ளனர்."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "மருத்துவ காப்புறுதி மூலம் 75,000.00 ரூபாய்கள் மீளப் பெறப்பட்டது."}, {"role": "assistant", "content": "மருத்துவ காப்புறுதி மூலம் எழுபத்தாயிரத்து ஐந்நூறு ரூபாய் பூஜ்ஜியம் பூஜ்ஜியம் ரூபாய்கள் மீளப் பெறப்பட்டது."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "சிறப்பு offer மூலம் 25% டிஸ்கவுண்ட் பெறலாம்."}, {"role": "assistant", "content": "சிறப்பு offer மூலம் இருபத்தைந்து சதவீத டிஸ்கவுண்ட் பெறலாம்."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "எனது ஆர்டர் நம்பர் 8-2-3-4-5-6-7-8-9, இன்னும் டெலிவரி ஆகவில்லை."}, {"role": "assistant", "content": "எனது ஆர்டர் நம்பர் எட்டு இரண்டு மூன்று நான்கு ஐந்து ஆறு ஏழு எட்டு ஒன்பது, இன்னும் டெலிவரி ஆகவில்லை."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "உங்களுடைய அக்ரவுண்ட் பேலன்ஸ் ரூபா 150.50 சதம் உள்ளது."}, {"role": "assistant", "content": "உங்களுடைய அக்ரவுண்ட் பேலன்ஸ் ரூபா நூற்று ஐம்பது ரூபாய் ஐம்பது சதம் உள்ளது."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இலவச டேட்டா பேக் 2024 ஆகஸ்ட் 15 வரை செல்லுபடியாகும்."}, {"role": "assistant", "content": "இலவச டேட்டா பேக் இரண்டாயிரத்து இருபத்தி நான்கு ஆகஸ்ட் பதினைந்து வரை செல்லுபடியாகும்."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "நெட்வொர்க் பராமரிப்பு காரணமாக இன்று இரவு 12:00am முதல் சேவை தடைப்படும்."}, {"role": "assistant", "content": "நெட்வொர்க் பராமரிப்பு காரணமாக இன்று இரவு பன்னிரண்டு மணி முதல் சேவை தடைப்படும்."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "இந்த திட்டத்தின் 1st phase வேலைகள் நிறைவடைந்துவிட்டன."}, {"role": "assistant", "content": "இந்த திட்டத்தின் முதலாம் phase வேலைகள் நிறைவடைந்துவிட்டன."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "எனது தொலைபேசி கணக்கிற்கு ரூ. 1,200 செலுத்தியதற்கான Transaction ID: 882910-Z ஐ இணைத்துள்ளேன்."}, {"role": "assistant", "content": "எனது தொலைபேசி கணக்கிற்கு ஒரு ஆயிரத்து இருநூறு ரூபாய் செலுத்தியதற்கான Transaction ID: எட்டு எட்டு இரண்டு ஒன்பது ஒன்று பூஜ்ஜியம் hyphen Z ஐ இணைத்துள்ளேன்."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Tamil (Sri Lanka), ready to be synthesized. Tamil as used in Sri Lanka (Rs = ரூபாய், சதம்).\nRules:\n- Expand everything that is not speakable as written into Tamil (Sri Lanka) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Tamil (Sri Lanka) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "மருந்துப் prescription-இல் 1/2 tablet என்று குறிப்பிடப்பட்டுள்ளது."}, {"role": "assistant", "content": "மருந்துப் prescription-இல் அரை tablet என்று குறிப்பிடப்பட்டுள்ளது."}], "lang": "ta-LK", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමට රු. 25,000.00 ක මුදලක් බැර කර ඇත."}, {"role": "assistant", "content": "ඔබේ ගිණුමට රුපියල් විසිපන්දහසක මුදලක් බැර කර ඇත."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ රක්ෂණ ආවරණය රු. 2,500,000 දක්වා වැඩි කරගත හැක."}, {"role": "assistant", "content": "ඔබේ රක්ෂණ ආවරණය රුපියල් විසිපන් ලක්ෂයක් දක්වා වැඩි කරගත හැක."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Emergency assistance සඳහා +94 11 234 5678 අමතන්න."}, {"role": "assistant", "content": "Emergency assistance සඳහා ප්ලස් නව හතර එක එක දෙ තුන හතර පහ හය හත අට අමතන්න."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "කරුණාකර ඔබේ PIN අංකය 5631 රහසිගතව තබා ගන්න."}, {"role": "assistant", "content": "කරුණාකර ඔබේ පින් අංකය පහ හය තුන එක රහසිගතව තබා ගන්න."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "ඔබේ ගිණුමෙන් රු. 5000-10000 අතර මුදලක් ලබාගැනීමට ATM යන්ත්‍රය භාවිතා කරන්න."}, {"role": "assistant", "content": "ඔබේ ගිණුමෙන් රුපියල් පන්දහසත් දසදහසත් අතර මුදලක් ලබාගැනීමට ඒ.ටී.එම් යන්ත්‍රය භාවිතා කරන්න."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "අලුත් data package එකක් activate කරගත්තොත් දින 7-14 දක්වා වලංගු කාලයක් ලැබෙනවා."}, {"role": "assistant", "content": "අලුත් data package එකක් activate කරගත්තොත් දින හතේ සිට දාහතර දක්වා වලංගු කාලයක් ලැබෙනවා."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "අද දින උෂ්ණත්වය 32°C පමණ වන නිසා වැඩිපුර ජලය පානය කරන්න."}, {"role": "assistant", "content": "අද දින උෂ්ණත්වය සෙල්සියස් අංශක තිස්දෙකක් පමණ වන නිසා වැඩිපුර ජලය පානය කරන්න."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "දැනට ඔබේ balance එක රු. 450.50ක් වන අතර data pack එකේ ID එක DP2024 වේ."}, {"role": "assistant", "content": "දැනට ඔබේ balance එක රුපියල් හාරසිය පනහයි දශම පනහක් වන අතර data pack එකේ ID එක ඩී පී දෙබින්ස් හතර වේ."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Sinhala, ready to be synthesized. Sinhala (රු. = රුපියල්).\nRules:\n- Expand everything that is not speakable as written into Sinhala words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Sinhala even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "රෝගියාගේ Glucose මට්ටම 110mg/dL වන අතර එය 5% කින් අඩු වී ඇත."}, {"role": "assistant", "content": "රෝගියාගේ Glucose මට්ටම එකසිය දහය මිලිග්‍රෑම් පර් ඩෙසිලීටර් වන අතර එය සියයට පහකින් අඩු වී ඇත."}], "lang": "si", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kailangan mong magpahinga nang husto para gumaling agad."}, {"role": "assistant", "content": "Kailangan mong magpahinga nang husto para gumaling agad."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mag-load po kayo ng ₱50 para ma-activate ang registration ng SIM."}, {"role": "assistant", "content": "Mag-load po kayo ng limampung piso para ma-activate ang registration ng SIM."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang interest rate ng savings account ko ay 0.25% lang kada taon."}, {"role": "assistant", "content": "Ang interest rate ng savings account ko ay zero punto dalawampu't lima porsyento lang kada taon."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Nag-drop ang signal strength sa 2.4GHz kaya mabagal ang connection."}, {"role": "assistant", "content": "Nag-drop ang signal strength sa dalawa punto apat gigahertz kaya mabagal ang connection."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "May 25% discount ang monthly plan mo kung mag-eenroll ka sa autopay."}, {"role": "assistant", "content": "May dalawampu't limang porsyentong discount ang monthly plan mo kung mag-eenroll ka sa autopay."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para sa mga inquiries tungkol sa loan, tumawag lang sa (02) 8123-4567 mula Lunes hanggang Biyernes."}, {"role": "assistant", "content": "Para sa mga inquiries tungkol sa loan, tumawag lang sa zero dalawa walo isa dalawa tatlo apat lima anim pito mula Lunes hanggang Biyernes."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang iyong credit card statement para sa Oktubre 2023 ay available na sa mobile app."}, {"role": "assistant", "content": "Ang iyong credit card statement para sa Oktubre dalawang libo dalawampu't tatlo ay available na sa mobile app."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kailangan ng medical history mula 1998 hanggang ngayon para sa clearance."}, {"role": "assistant", "content": "Kailangan ng medical history mula isang libo siyam na raan at siyamnapu't walo hanggang ngayon para sa clearance."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para sa inyong medical records, paki-email lang kami sa records_dept@cityhealth.ph."}, {"role": "assistant", "content": "Para sa inyong medical records, paki-email lang kami sa records underscore dept at cityhealth dot ph."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Pa-verify naman po ng code na 482910 para ma-transfer ko na yung ₱5,000 sa account mo today."}, {"role": "assistant", "content": "Pa-verify naman po ng code na apat walo dalawa siyam isa zero para ma-transfer ko na yung limang libong piso sa account mo today."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ang schedule ng check-up ay tuwing 1/2 ng buwan."}, {"role": "assistant", "content": "Ang schedule ng check-up ay tuwing kalahati ng buwan."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Filipino (Tagalog), ready to be synthesized. Filipino/Tagalog (₱ = piso); Taglish is acceptable.\nRules:\n- Expand everything that is not speakable as written into Filipino (Tagalog) words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Filipino (Tagalog) even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kailangan itabi itong sample sa -80°C para hindi masira."}, {"role": "assistant", "content": "Kailangan itabi itong sample sa negative walumpung digri Celsius para hindi masira."}], "lang": "tl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "تم إيداع مبلغ 150.75 درهم في المحفظة الإلكترونية."}, {"role": "assistant", "content": "تم إيداع مبلغ مائة وخمسون فاصلة خمسة وسبعون درهماً في المحفظة الإلكترونية."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "قيمة التبرعات الواردة لبناء المركز الصحي بلغت 5 000 000 جنيه"}, {"role": "assistant", "content": "قيمة التبرعات الواردة لبناء المركز الصحي بلغت خمسة ملايين جنيه"}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "نحيطكم علماً بأن رصيدكم الحالي هو 200 جنيه، لمزيد من التفاصيل اتصل بـ 01012345678"}, {"role": "assistant", "content": "نحيطكم علماً بأن رصيدكم الحالي هو مائتان جنيه، لمزيد من التفاصيل اتصل بـ صفر واحد صفر واحد اثنان ثلاثة أربعة خمسة ستة سبعة ثمانية"}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى تزويدنا بالرمز البريدي 1 2 4 5 3 لتحديد موقع أقرب صيدلية"}, {"role": "assistant", "content": "يرجى تزويدنا بالرمز البريدي واحد اثنان أربعة خمسة ثلاثة لتحديد موقع أقرب صيدلية"}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى مراجعة فرع البنك لتحديث بيانات هويتك قبل 2024/12/15 لتجنب تجميد الحساب."}, {"role": "assistant", "content": "يرجى مراجعة فرع البنك لتحديث بيانات هويتك قبل الخامس عشر من ديسمبر ألفين وأربعة وعشرين لتجنب تجميد الحساب."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى إحضار نتائج التحاليل في زيارتك القادمة بتاريخ 22/09."}, {"role": "assistant", "content": "يرجى إحضار نتائج التحاليل في زيارتك القادمة بتاريخ الثاني والعشرين من سبتمبر."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "يرجى مراجعة مدير الحساب في تمام الـ 12:00 ظهراً لتوقيع الأوراق"}, {"role": "assistant", "content": "يرجى مراجعة مدير الحساب في تمام الـ الثانية عشرة تماماً ظهراً لتوقيع الأوراق"}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "المسافة بين الفرع الرئيسي وأقرب صراف آلي هي 1.5 km تقريباً."}, {"role": "assistant", "content": "المسافة بين الفرع الرئيسي وأقرب صراف آلي هي واحد فاصلة خمسة كيلومتر تقريباً."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "تؤكد التقارير الطبية أن المريض يعاني من هذه الحالة منذ عام 1992"}, {"role": "assistant", "content": "تؤكد التقارير الطبية أن المريض يعاني من هذه الحالة منذ عام ألف وتسعمائة واثنين وتسعين"}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "عزيزي العميل، تم تفعيل باقة الإنترنت الخاصة بك برقم الاشتراك 966501234567."}, {"role": "assistant", "content": "عزيزي العميل، تم تفعيل باقة الإنترنت الخاصة بك برقم الاشتراك تسعة ستة ستة خمسة صفر واحد اثنان ثلاثة أربعة خمسة ستة سبعة."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "تم سداد مبلغ 2,300 درهم بتاريخ 12-08-2023 عبر التطبيق."}, {"role": "assistant", "content": "تم سداد مبلغ ألفين وثلاثمائة درهم بتاريخ اثنا عشر ثمانية ألفين وثلاثة وعشرين عبر التطبيق."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Arabic, ready to be synthesized. Modern Standard Arabic; amounts may be in riyal, dirham, pound or dollar.\nRules:\n- Expand everything that is not speakable as written into Arabic words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Arabic even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "نسبة الفائدة السنوية هي 4.25%، والمبلغ المطلوب سداده هو 10,000 جنيه."}, {"role": "assistant", "content": "نسبة الفائدة السنوية هي أربعة فاصلة خمسة وعشرون بالمئة، والمبلغ المطلوب سداده هو عشرة آلاف جنيه."}], "lang": "ar", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Je rencontre des problèmes de connexion avec ma box internet."}, {"role": "assistant", "content": "Je rencontre des problèmes de connexion avec ma box internet."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il reste 3 jours pour contester ce prélèvement."}, {"role": "assistant", "content": "Il reste trois jours pour contester ce prélèvement."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le coût annuel des soins pour cette pathologie s'élève à 18 000 € par patient."}, {"role": "assistant", "content": "Le coût annuel des soins pour cette pathologie s'élève à dix-huit mille euros par patient."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Je n'arrive plus à me connecter à mon appli bancaire, pouvez-vous m'aider au 01 42 76 00 00 ?"}, {"role": "assistant", "content": "Je n'arrive plus à me connecter à mon appli bancaire, pouvez-vous m'aider au zéro un quatre deux sept six zéro zéro zéro zéro ?"}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Je n'arrive pas à valider ma commande n° 8302741, pouvez-vous m'aider ?"}, {"role": "assistant", "content": "Je n'arrive pas à valider ma commande numéro huit trois zéro deux sept quatre un, pouvez-vous m'aider ?"}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Votre commande n° FR-910283 a été expédiée ce matin."}, {"role": "assistant", "content": "Votre commande numéro F R tiret neuf un zéro deux huit trois a été expédiée ce matin."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le rendez-vous avec le cardiologue est fixé au lundi 15 mai à 14h."}, {"role": "assistant", "content": "Le rendez-vous avec le cardiologue est fixé au lundi quinze mai à quatorze heures."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'offre promotionnelle telco se termine le 30 sept."}, {"role": "assistant", "content": "L'offre promotionnelle telco se termine le trente septembre."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Merci de confirmer votre présence à l'examen du 22 juin 2023."}, {"role": "assistant", "content": "Merci de confirmer votre présence à l'examen du vingt-deux juin deux mille vingt-trois."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le virement sera effectif demain à 08h."}, {"role": "assistant", "content": "Le virement sera effectif demain à huit heures."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "La dose recommandée est de 2-3 gouttes dans chaque œil."}, {"role": "assistant", "content": "La dose recommandée est de deux à trois gouttes dans chaque œil."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le virement de 1500€ a bien été enregistré sur votre compte courant."}, {"role": "assistant", "content": "Le virement de mille cinq cents euros a bien été enregistré sur votre compte courant."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Facture mobile du mois : 24,99€. Détail sur https://www.free.fr/facturation"}, {"role": "assistant", "content": "Facture mobile du mois : vingt-quatre euros quatre-vingt-dix-neuf. Détail sur hachettage double slash www point free point fr slash facturation"}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in French, ready to be synthesized. French (France).\nRules:\n- Expand everything that is not speakable as written into French words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in French even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Le taux d'intérêt annuel est passé à 3,1% le 01/01/2024."}, {"role": "assistant", "content": "Le taux d'intérêt annuel est passé à trois virgule un pour cent le premier janvier deux mille vingt-quatre."}], "lang": "fr", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El banco ha decidido actualizar las condiciones de las comisiones de mantenimiento."}, {"role": "assistant", "content": "El banco ha decidido actualizar las condiciones de las comisiones de mantenimiento."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El presupuesto asignado para el nuevo ala del hospital es de 12.000.000 €."}, {"role": "assistant", "content": "El presupuesto asignado para el nuevo ala del hospital es de doce millones de euros."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Te informamos que hemos aplicado un descuento del 12,5% en tu última factura de fibra."}, {"role": "assistant", "content": "Te informamos que hemos aplicado un descuento del doce coma cinco por ciento en tu última factura de fibra."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "La eficacia de la vacuna es del 92,5% según el último estudio clínico."}, {"role": "assistant", "content": "La eficacia de la vacuna es del noventa y dos coma cinco por ciento según el último estudio clínico."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Tiene una incidencia abierta en su línea móvil desde el 14 de sept."}, {"role": "assistant", "content": "Tiene una incidencia abierta en su línea móvil desde el catorce de sept."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "La promoción de fibra termina hoy a las 12 pm."}, {"role": "assistant", "content": "La promoción de fibra termina hoy a las doce pm."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "La transferencia se ejecutará automáticamente a las 00:00."}, {"role": "assistant", "content": "La transferencia se ejecutará automáticamente a las cero cero cero cero."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El paciente presenta una fiebre de 38-39ºC."}, {"role": "assistant", "content": "El paciente presenta una fiebre de treinta y ocho a treinta y nueve grados Celsius."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para el análisis de sangre, por favor, traiga el bote con 10 ml de muestra."}, {"role": "assistant", "content": "Para el análisis de sangre, por favor, traiga el bote con diez mililitros de muestra."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "El medicamento debe administrarse en dosis de 5 mg cada 8 horas."}, {"role": "assistant", "content": "El medicamento debe administrarse en dosis de cinco miligramos cada ocho horas."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Para gestionar el roaming en su viaje a USA, mande un correo a viajes@telecom.es."}, {"role": "assistant", "content": "Para gestionar el roaming en su viaje a USA, mande un correo a viajes arroba telecom punto es."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Me gustaría reclamar el cargo de 45€ asociado al recibo Bnk-00129."}, {"role": "assistant", "content": "Me gustaría reclamar el cargo de cuarenta y cinco euros asociado al recibo Bnk guion cero cero uno dos nueve."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Spanish, ready to be synthesized. Spanish (Spain).\nRules:\n- Expand everything that is not speakable as written into Spanish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Spanish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Llevamos tres días con temperaturas extremas de hasta -4ºC en el interior."}, {"role": "assistant", "content": "Llevamos tres días con temperaturas extremas de hasta menos cuatro grados Celsius en el interior."}], "lang": "es", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mein Mobilfunkvertrag läuft am Ende des Monats aus."}, {"role": "assistant", "content": "Mein Mobilfunkvertrag läuft am Ende des Monats aus."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mein Handy-Tarif kostet monatlich 19,99 €."}, {"role": "assistant", "content": "Mein Handy-Tarif kostet monatlich neunzehn Euro neunundneunzig."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Anschlussgebühr von 39,90 EUR wird mit der ersten Rechnung verrechnet."}, {"role": "assistant", "content": "Die Anschlussgebühr von neununddreißig Euro neunzig wird mit der ersten Rechnung verrechnet."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Wir haben Ihr Paket auf 100 Mbit/s gedrosselt, da Sie das Limit von 1,5 TB überschritten haben."}, {"role": "assistant", "content": "Wir haben Ihr Paket auf einhundert Megabit pro Sekunde gedrosselt, da Sie das Limit von eins Komma fünf Terabyte überschritten haben."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte beachten Sie, dass die Mehrwertsteuer von 19% bereits im Preis enthalten ist."}, {"role": "assistant", "content": "Bitte beachten Sie, dass die Mehrwertsteuer von neunzehn Prozent bereits im Preis enthalten ist."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ein Anteil von 12.5% der Patienten zeigte leichte Nebenwirkungen."}, {"role": "assistant", "content": "Ein Anteil von zwölf Komma fünf Prozent der Patienten zeigte leichte Nebenwirkungen."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Schicken Sie mir die Bestätigung bitte per SMS an 0152 4433221."}, {"role": "assistant", "content": "Schicken Sie mir die Bestätigung bitte per SMS an null eins fünf zwei vier vier drei drei zwei zwei eins."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Bitte geben Sie Ihre 4-stellige PIN am Geldautomaten ein."}, {"role": "assistant", "content": "Bitte geben Sie Ihre vierstellige PIN am Geldautomaten ein."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Überweisung auf das Konto DE89 3704 0044 0532 1190 00 ist leider fehlgeschlagen."}, {"role": "assistant", "content": "Die Überweisung auf das Konto D E acht neun drei sieben null vier null null vier vier null fünf drei zwei eins eins neun null null null ist leider fehlgeschlagen."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Für Ihren Termin in der Praxis benötigen wir Ihr Versicherungsnummer-Kürzel, zum Beispiel A123456789."}, {"role": "assistant", "content": "Für Ihren Termin in der Praxis benötigen wir Ihr Versicherungsnummer-Kürzel, zum Beispiel A eins zwei drei vier fünf sechs sieben acht neun."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ihr Vertrag läuft am 30.06. aus, falls Sie nicht kündigen."}, {"role": "assistant", "content": "Ihr Vertrag läuft am dreißigsten Juni aus, falls Sie nicht kündigen."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ihr Termin findet im Zeitraum von 14-15 Uhr statt."}, {"role": "assistant", "content": "Ihr Termin findet im Zeitraum von vierzehn bis fünfzehn Uhr statt."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Vertragslaufzeit beträgt je nach Modell 12-24 Monate."}, {"role": "assistant", "content": "Die Vertragslaufzeit beträgt je nach Modell zwölf bis vierundzwanzig Monate."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Für weitere Infos zu Ihrem Tarif schauen Sie doch mal auf www.telekom.de vorbei."}, {"role": "assistant", "content": "Für weitere Infos zu Ihrem Tarif schauen Sie doch mal auf w w w Punkt telekom Punkt de vorbei."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Melden Sie sich bitte via https://kundenservice.vodafone.de/login an, um Ihren Vertrag zu ändern."}, {"role": "assistant", "content": "Melden Sie sich bitte via h t t p s Doppelpunkt Schrägstrich Schrägstrich kundenservice Punkt vodafone Punkt de Schrägstrich login an, um Ihren Vertrag zu ändern."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in German, ready to be synthesized. German (Germany).\nRules:\n- Expand everything that is not speakable as written into German words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in German even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Die Verfügbarkeit unseres Netzes lag im letzten Quartal bei 99,9/100 %."}, {"role": "assistant", "content": "Die Verfügbarkeit unseres Netzes lag im letzten Quartal bei neunundneunzig Komma neun von einhundert Prozent."}], "lang": "de", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il costo della visita domiciliare è di circa 60€."}, {"role": "assistant", "content": "Il costo della visita domiciliare è di circa sessanta euro."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il bonus di benvenuto di 50€ verrà accreditato entro 48 ore."}, {"role": "assistant", "content": "Il bonus di benvenuto di cinquanta euro verrà accreditato entro quarantotto ore."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "L'offerta prevede uno sconto del 15,5% sul canone mensile."}, {"role": "assistant", "content": "L'offerta prevede uno sconto del quindici virgola cinque per cento sul canone mensile."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tasso di saturazione dell'ossigeno è sceso al 25% sotto il livello normale."}, {"role": "assistant", "content": "Il tasso di saturazione dell'ossigeno è sceso al venticinque per cento sotto il livello normale."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Per l'attivazione della SIM, invia un SMS al 190 o chiama il +39 333 1234567."}, {"role": "assistant", "content": "Per l'attivazione della SIM, invia un SMS al uno nove zero o chiama il più trentanove tre tre tre uno due tre quattro cinque sei sette."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Potrebbe dettarmi il numero di conto? Sì, è 8-0-1-2-4-5-6-7-8."}, {"role": "assistant", "content": "Potrebbe dettarmi il numero di conto? Sì, è otto zero uno due quattro cinque sei sette otto."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il codice di verifica per il reset password è 0-9-3-7."}, {"role": "assistant", "content": "Il codice di verifica per il reset password è zero nove tre sette."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Prenotazione fissata per il 14 gen 2024 alle ore 10:00."}, {"role": "assistant", "content": "Prenotazione fissata per il quattordici gennaio duemilaventiquattro alle ore dieci zero zero."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Ho chiamato l'assistenza per la 3ª volta oggi e non ho ancora risolto."}, {"role": "assistant", "content": "Ho chiamato l'assistenza per la terza volta oggi e non ho ancora risolto."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il 10° paziente della lista d'attesa è stato appena chiamato."}, {"role": "assistant", "content": "Il decimo paziente della lista d'attesa è stato appena chiamato."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il paziente deve restare a digiuno per 8-12 ore prima del prelievo."}, {"role": "assistant", "content": "Il paziente deve restare a digiuno per otto a dodici ore prima del prelievo."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tasso d'interesse aggiornato a gennaio 2024 è del 3,5%."}, {"role": "assistant", "content": "Il tasso d'interesse aggiornato a gennaio duemilaventiquattro è del tre virgola cinque per cento."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Negli anni '70 la prevenzione cardiaca era molto meno avanzata."}, {"role": "assistant", "content": "Negli anni settanta la prevenzione cardiaca era molto meno avanzata."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Se desidera chiudere il conto, mandi una PEC a ufficioperativi@banca-italia.it."}, {"role": "assistant", "content": "Se desidera chiudere il conto, mandi una PEC a ufficioperativi chiocciola banca trattino italia punto it."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il tuo abbonamento scade il 20/02/2025 alle ore 23:59."}, {"role": "assistant", "content": "Il tuo abbonamento scade il venti febbraio duemilaventicinque alle ore ventitré e cinquantanove."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Italian, ready to be synthesized. Italian.\nRules:\n- Expand everything that is not speakable as written into Italian words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Italian even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Il segnale 4G/5G in questa zona è purtroppo molto instabile."}, {"role": "assistant", "content": "Il segnale quattro G slash cinque G in questa zona è purtroppo molto instabile."}], "lang": "it", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O paciente apresenta uma tensão arterial de 130/85."}, {"role": "assistant", "content": "O paciente apresenta uma tensão arterial de cento e trinta por oitenta e cinco."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A fatura do telemóvel este mês veio com um erro de 15,90€."}, {"role": "assistant", "content": "A fatura do telemóvel este mês veio com um erro de quinze euros e noventa cêntimos."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O saldo da minha conta corrente é de 1.250,45€."}, {"role": "assistant", "content": "O saldo da minha conta corrente é de mil duzentos e cinquenta euros e quarenta e cinco cêntimos."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A mensalidade do plano de dados é de 24,99€."}, {"role": "assistant", "content": "A mensalidade do plano de dados é de vinte e quatro euros e noventa e nove cêntimos."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A minha internet está 100% lenta hoje, preciso de suporte técnico."}, {"role": "assistant", "content": "A minha internet está cem por cento lenta hoje, preciso de suporte técnico."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Confirmamos a encomenda n.º 8 3 0 5 7 2; o envio será feito amanhã."}, {"role": "assistant", "content": "Confirmamos a encomenda número oito três zero cinco sete dois; o envio será feito amanhã."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O enfermeiro disse que a medicação deve ser administrada às 15:00 e às 21:00."}, {"role": "assistant", "content": "O enfermeiro disse que a medicação deve ser administrada às quinze horas e às vinte e uma horas."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A 1ª transferência de 500€ foi processada com sucesso."}, {"role": "assistant", "content": "A primeira transferência de quinhentos euros foi processada com sucesso."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A cobertura de rede 5G está disponível em 80-90% do território continental."}, {"role": "assistant", "content": "A cobertura de rede cinco gé está disponível em oitenta a noventa por cento do território continental."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O tempo de espera na urgência pode variar entre 2-6 horas."}, {"role": "assistant", "content": "O tempo de espera na urgência pode variar entre duas a seis horas."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Estou à espera dos exames que pedi no início dos anos 2000."}, {"role": "assistant", "content": "Estou à espera dos exames que pedi no início dos anos dois mil."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O relatório de alta hospitalar foi encaminhado para o médico assistente em dr.costa@clinicacentro.pt."}, {"role": "assistant", "content": "O relatório de alta hospitalar foi encaminhado para o médico assistente em dê érre ponto costa arroba clínica centro ponto pt."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "A consulta do utente com o n.º NHS-55432 está marcada para amanhã."}, {"role": "assistant", "content": "A consulta do utente com o número NHS cinco cinco quatro três dois está marcada para amanhã."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Por favor, transfira os 150,50€ para o IBAN PT50 0001 0000 1234 5678 9012 3."}, {"role": "assistant", "content": "Por favor, transfira os cento e cinquenta euros e cinquenta cêntimos para o IBAN Pê Tê cinco zero zero zero zero um zero zero zero zero um dois três quatro cinco seis sete oito nove zero um dois três."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O carregamento de 20€ para o número 912345678 foi efetuado com sucesso."}, {"role": "assistant", "content": "O carregamento de vinte euros para o número nove um dois três quatro cinco seis sete oito foi efetuado com sucesso."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O juro do crédito habitação subiu 0,25% e a próxima prestação vence a 01/12."}, {"role": "assistant", "content": "O juro do crédito habitação subiu zero vírgula vinte e cinco por cento e a próxima prestação vence a zero um de dezembro."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Portuguese, ready to be synthesized. European Portuguese (also R$ real).\nRules:\n- Expand everything that is not speakable as written into Portuguese words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Portuguese even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "O paciente deve tomar 1/2 comprimido de 500mg antes das refeições."}, {"role": "assistant", "content": "O paciente deve tomar meio comprimido de quinhentos miligramas antes das refeições."}], "lang": "pt", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mijn internetverbinding valt sinds gisteren regelmatig weg."}, {"role": "assistant", "content": "Mijn internetverbinding valt sinds gisteren regelmatig weg."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Er zijn ongeveer 1.100.000 mensen geregistreerd in dit specifieke zorgprogramma."}, {"role": "assistant", "content": "Er zijn ongeveer één miljoen honderdduizend mensen geregistreerd in dit specifieke zorgprogramma."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De huidige rentevoet voor een hypotheek van 10 jaar is ongeveer 3,85%."}, {"role": "assistant", "content": "De huidige rentevoet voor een hypotheek van tien jaar is ongeveer drie komma vijf en tachtig procent."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "De effectiviteit van dit medicijn ligt rond de 85-90%."}, {"role": "assistant", "content": "De effectiviteit van dit medicijn ligt rond de vijfentachtig tot negentig procent."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Kunt u morgen om 14:30 uur langskomen voor het tekenen van de hypotheekakte?"}, {"role": "assistant", "content": "Kunt u morgen om veertien uur dertig minuten langskomen voor het tekenen van de hypotheekakte?"}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Mijn simkaart is geactiveerd om 9 am vanochtend."}, {"role": "assistant", "content": "Mijn simkaart is geactiveerd om negen uur vanochtend."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Voor meer info over je mobiele abonnement ga je naar www.kpn.nl/service."}, {"role": "assistant", "content": "Voor meer info over je mobiele abonnement ga je naar www punt kpn punt nl slash service."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "We garanderen een uptime van 99,9/100 voor al onze zakelijke lijnen."}, {"role": "assistant", "content": "We garanderen een uptime van negenennegentig komma negen op honderd voor al onze zakelijke lijnen."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Dutch, ready to be synthesized. Dutch (Netherlands).\nRules:\n- Expand everything that is not speakable as written into Dutch words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Dutch even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Let op: bij temperaturen onder -10°C kan de batterijduur van uw toestel afnemen."}, {"role": "assistant", "content": "Let op: bij temperaturen onder min tien graden Celsius kan de batterijduur van uw toestel afnemen."}], "lang": "nl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoje saldo wynosi obecnie 1240,50 zł."}, {"role": "assistant", "content": "Twoje saldo wynosi obecnie tysiąc dwieście czterdzieści przecinek pięć zero złotych."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Skuteczność tego leku w grupie kontrolnej wyniosła 88,5%, co jest wynikiem bardzo dobrym."}, {"role": "assistant", "content": "Skuteczność tego leku w grupie kontrolnej wyniosła osiemdziesiąt osiem przecinek pięć procent, co jest wynikiem bardzo dobrym."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Proszę podać numer telefonu do kontaktu w razie nagłych wypadków, np. 600 000 000."}, {"role": "assistant", "content": "Proszę podać numer telefonu do kontaktu w razie nagłych wypadków, np. sześć zero zero zero zero zero zero zero zero zero."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Moim numerem konta jest PL 49 1020 1013 0000 0102 0345 6789."}, {"role": "assistant", "content": "Moim numerem konta jest P L cztery dziewięć jeden zero dwa zero jeden zero jeden trzy zero zero zero zero zero jeden zero dwa zero trzy cztery pięć sześć siedem osiem dziewięć."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Przypominamy o konieczności odbioru recepty do 30 czerwca."}, {"role": "assistant", "content": "Przypominamy o konieczności odbioru recepty do trzydziestego czerwca."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Proszę zapisać się na 12. piętro do gabinetu okulistycznego."}, {"role": "assistant", "content": "Proszę zapisać się na dwunaste piętro do gabinetu okulistycznego."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Oferta promocyjna obowiązuje dla klientów w wieku 18-26 lat."}, {"role": "assistant", "content": "Oferta promocyjna obowiązuje dla klientów w wieku osiemnastu do dwudziestu sześciu lat."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twój pakiet danych obejmuje 50 GB szybkiego internetu LTE."}, {"role": "assistant", "content": "Twój pakiet danych obejmuje pięćdziesiąt gigabajtów szybkiego internetu LTE."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Badania krwi z 2023 r. wskazują na znaczną poprawę wyników."}, {"role": "assistant", "content": "Badania krwi z dwa tysiące dwudziestego trzeciego roku wskazują na znaczną poprawę wyników."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Szczegóły Twojej nowej taryfy są dostępne tutaj: www.telco-plus.pl/mojekonto"}, {"role": "assistant", "content": "Szczegóły Twojej nowej taryfy są dostępne tutaj: w w w kropka telco kreska plus kropka pl ukośnik moje konto"}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Proszę o pilną weryfikację przelewu o numerze referencyjnym PL9820001234567890."}, {"role": "assistant", "content": "Proszę o pilną weryfikację przelewu o numerze referencyjnym P L dziewięć osiem dwa zero zero zero jeden dwa trzy cztery pięć sześć siedem osiem dziewięć zero."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Proszę sprawdzić status wizyty w systemie, wpisując kod potwierdzenia VIZ-8821."}, {"role": "assistant", "content": "Proszę sprawdzić status wizyty w systemie, wpisując kod potwierdzenia V I Z osiem osiem dwa jeden."}], "lang": "pl", "source": "llm"} {"messages": [{"role": "system", "content": "You are the text normalizer of a multilingual text-to-speech (TTS) system. Rewrite the user's text into its exact spoken form in Polish, ready to be synthesized. Polish.\nRules:\n- Expand everything that is not speakable as written into Polish words, the way a native speaker reads the sentence aloud: numbers, money (say the currency), phone numbers digit by digit, ID/reference/OTP/postcode numbers digit by digit, dates, times, percentages, decimals, fractions, ordinals, units, email addresses and URLs.\n- Verbalize in Polish even where a Latin acronym or name sits next to the number. Never translate the text itself.\n- Never answer questions, never add, drop, reorder or translate words. Only rewrite non-speakable tokens into words; everything else stays exactly as written, including punctuation.\n- If nothing needs normalizing, return the text unchanged.\n- Reply with JSON only: {\"normalized\": \"\"}"}, {"role": "user", "content": "Twoja faktura za telefon jest opłacona w 2/3."}, {"role": "assistant", "content": "Twoja faktura za telefon jest opłacona w dwóch trzecich."}], "lang": "pl", "source": "llm"}