{
"name": "02_yoruba_greetings_english",
"desc": "Yoruba greetings + English. User turns TTS-synthesized (Flash, female Yoruba voice).",
"user_turns": [
{
"label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"text": "Ẹ káàárọ̀! I want to buy a new phone for my sister.",
"wav": "/scratch/duplex_v2/user_audio/02_yoruba_greetings_english_turn0.wav",
"orig_label": "TTS-synthesized user turn"
},
{
"label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"text": "Her budget is about one hundred thousand naira.",
"wav": "/scratch/duplex_v2/user_audio/02_yoruba_greetings_english_turn1.wav",
"orig_label": "TTS-synthesized user turn"
},
{
"label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"text": "Okay, that sounds good. Ẹ ṣé!",
"wav": "/scratch/duplex_v2/user_audio/02_yoruba_greetings_english_turn2.wav",
"orig_label": "TTS-synthesized user turn"
}
],
"turns": [
{
"turn": 0,
"user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"user_ref": "Ẹ káàárọ̀! I want to buy a new phone for my sister.",
"user_start": 0.7,
"user_onset": 0.715,
"backchannels": [],
"asr": "ẹka arọ i want to buy any phone for my sisters",
"asr_s": 0.079,
"eot_wait_s": 0.576,
"speech_end": 4.096,
"target": {
"mix": {
"eng": 0.82,
"yor": 0.18
},
"matrix": "eng",
"pattern": [
"inter-sentential"
]
},
"target_desc": "Match the user's language mix: English about 80%, Yoruba about 20% of the words. The main (matrix) language is English.",
"reply": "Okay, I hear you. What kind of phone are you thinking for your sisters? Ọmọ rẹ wà?",
"reply_mix": {
"eng": 0.82,
"yor": 0.18
},
"tvd": 0.002,
"how": "first_round",
"first_cand_ok": true,
"rounds": [
{
"round": 0,
"llm_s": 0.851,
"cands": [
{
"text": "Okay, I understand. What kind of phone are you looking for—new or used? Àwọn àwọn àwọn àwọn?",
"ok": true,
"why": [],
"tvd": 0.048
},
{
"text": "Okay, I hear you. What kind of phone are you thinking for your sisters? Ọmọ rẹ wà?",
"ok": true,
"why": [],
"tvd": 0.002
},
{
"text": "Ẹ káàárọ̀! What kind of phone are you looking for—brand, price, or features? Ẹ káàárọ̀!",
"ok": true,
"why": [],
"tvd": 0.012
}
]
}
],
"mix_s": 0.0076,
"llm_s": 0.851,
"tts_first_s": 1.032,
"spec_start_s": 0.256,
"compute_to_first_audio_s": 1.984,
"tts_breakdown": {
"fe": 0.0,
"lm": 0.616,
"flow_voc": 0.22,
"n_tok": 39,
"audio_s": 1.56
},
"tts_chunks_s": [
1.032,
1.483,
0.836
],
"first_audio": 6.335,
"reply_end": 12.895,
"latency_s": 2.239
},
{
"turn": 1,
"user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"user_ref": "Her budget is about one hundred thousand naira.",
"user_start": 13.595,
"user_onset": 13.83,
"backchannels": [],
"asr": "her budget is about 100 000 naira",
"asr_s": 0.066,
"eot_wait_s": 0.576,
"speech_end": 16.736,
"target": {
"mix": {
"eng": 0.95,
"yor": 0.05
},
"matrix": "eng",
"pattern": []
},
"target_desc": "Match the user's language mix: English about 95%, Yoruba about 5% of the words. The main (matrix) language is English.",
"reply": "Okay, with one hundred thousand naira, we can find a good smartphone for her. Any preferred brand? Ọmọ rẹ wà?",
"reply_mix": {
"eng": 0.85,
"yor": 0.15
},
"tvd": 0.1,
"how": "first_round",
"first_cand_ok": true,
"rounds": [
{
"round": 0,
"llm_s": 0.345,
"cands": [
{
"text": "Okay, with one hundred thousand naira, we can find a good smartphone for her. Any preferred brand? Ọmọ rẹ wà?",
"ok": true,
"why": [],
"tvd": 0.1
},
{
"text": "Okay, with one hundred thousand naira, we can find a good phone for her. Any brand preference? Ọmọ rẹ wà?",
"ok": true,
"why": [],
"tvd": 0.1
},
{
"text": "Okay, with a budget of one hundred thousand naira, I can suggest a few good phones. Do you want something with a good camera? Àwọn àkẹ́kẹ́ ní àwọn àwọn?",
"ok": true,
"why": [],
"tvd": 0.121
}
]
}
],
"mix_s": 0.0038,
"llm_s": 0.345,
"tts_first_s": 0.59,
"spec_start_s": 0.256,
"compute_to_first_audio_s": 1.021,
"tts_breakdown": {
"fe": 0.0,
"lm": 0.53,
"flow_voc": 0.221,
"n_tok": 34,
"audio_s": 1.36
},
"tts_chunks_s": [
0.59,
2.303,
1.014,
0.751
],
"first_audio": 18.013,
"reply_end": 28.357,
"latency_s": 1.277
},
{
"turn": 2,
"user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
"user_ref": "Okay, that sounds good. Ẹ ṣé!",
"user_start": 29.057,
"user_onset": 29.312,
"backchannels": [],
"asr": "oka that sounds good she",
"asr_s": 0.077,
"eot_wait_s": 0.576,
"speech_end": 31.808,
"target": {
"mix": {
"eng": 0.97
},
"matrix": "eng",
"pattern": []
},
"target_desc": "Match the user's language mix: English about 95% of the words. The main (matrix) language is English. Reply only in English.",
"reply": "Okay, that sounds good. She has a favorite brand in mind?",
"reply_mix": {
"eng": 1.0
},
"tvd": 0.015,
"how": "first_round",
"first_cand_ok": true,
"rounds": [
{
"round": 0,
"llm_s": 0.207,
"cands": [
{
"text": "Okay, that sounds good. She has a favorite brand in mind?",
"ok": true,
"why": [],
"tvd": 0.015
},
{
"text": "Okay, that sounds good. She likes any particular brand like Samsung, iPhone, or Tecno?",
"ok": true,
"why": [],
"tvd": 0.015
},
{
"text": "Great! She’ll love it. Should we look for a phone with a good camera or long battery life?",
"ok": true,
"why": [],
"tvd": 0.015
}
]
}
],
"mix_s": 0.0034,
"llm_s": 0.207,
"tts_first_s": 1.217,
"spec_start_s": 0.256,
"compute_to_first_audio_s": 1.515,
"tts_breakdown": {
"fe": 0.0,
"lm": 0.887,
"flow_voc": 0.221,
"n_tok": 55,
"audio_s": 2.2
},
"tts_chunks_s": [
1.217,
1.109
],
"first_audio": 33.579,
"reply_end": 38.219,
"latency_s": 1.771
}
]
}