samples/04_hausa_greetings_english.json

{
 "name": "04_hausa_greetings_english",
 "desc": "Hausa greeting + English, Hausa closing. User turns TTS-synthesized (Flash, female voice).",
 "user_turns": [
  {
   "label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
   "text": "<hau>Sannu, barka da yamma.</hau> <eng>I am travelling to Kano tomorrow.</eng>",
   "wav": "/scratch/duplex_v2/user_audio/04_hausa_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": "<eng>What should I pack for the weather there?</eng>",
   "wav": "/scratch/duplex_v2/user_audio/04_hausa_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": "<hau>To, na gode.</hau>",
   "wav": "/scratch/duplex_v2/user_audio/04_hausa_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": "<hau>Sannu, barka da yamma.</hau> <eng>I am travelling to Kano tomorrow.</eng>",
   "user_start": 0.7,
   "user_onset": 0.77,
   "backchannels": [],
   "asr": "sannu baka da yamma i am travelling to kano tomorrow",
   "asr_s": 0.088,
   "eot_wait_s": 0.576,
   "speech_end": 5.248,
   "target": {
    "mix": {
     "eng": 0.6,
     "hau": 0.4
    },
    "matrix": "eng",
    "pattern": [
     "greeting:hau",
     "discourse:hau",
     "inter-sentential"
    ]
   },
   "target_desc": "Match the user's language mix: English about 60%, Hausa about 40% of the words. The main (matrix) language is English. The user greeted in Hausa: open with a short Hausa greeting. Sprinkle a Hausa discourse word or two the way the user did.",
   "reply": "<hau>Sannu!</hau> <eng>Good evening, I hear you’re travelling to Kano tomorrow. Is it for business or pleasure?</eng> <hau>Ba a da kai yin?</hau>",
   "reply_mix": {
    "eng": 0.74,
    "hau": 0.26
   },
   "tvd": 0.137,
   "how": "first_round",
   "first_cand_ok": true,
   "rounds": [
    {
     "round": 0,
     "llm_s": 0.301,
     "cands": [
      {
       "text": "<hau>Sannu baka da yamma!</hau> <eng>Traveling to Kano tomorrow, eh? You need help with your flight or hotel booking?</eng>",
       "ok": true,
       "why": [],
       "tvd": 0.179
      },
      {
       "text": "<hau>Sannu!</hau> <eng>Good evening, I hear you're heading to Kano tomorrow. Any special reason for the trip?</eng> <hau>Baka da yamma</hau>",
       "ok": true,
       "why": [],
       "tvd": 0.19
      },
      {
       "text": "<hau>Sannu!</hau> <eng>Good evening, I hear you’re travelling to Kano tomorrow. Is it for business or pleasure?</eng> <hau>Ba a da kai yin?</hau>",
       "ok": true,
       "why": [],
       "tvd": 0.137
      }
     ]
    }
   ],
   "mix_s": 0.0076,
   "llm_s": 0.301,
   "tts_first_s": 0.679,
   "spec_start_s": 0.256,
   "compute_to_first_audio_s": 1.091,
   "tts_breakdown": {
    "fe": 0.0,
    "lm": 0.685,
    "flow_voc": 0.208,
    "n_tok": 43,
    "audio_s": 1.72
   },
   "tts_chunks_s": [
    0.679,
    1.738,
    1.186,
    0.894
   ],
   "first_audio": 6.595,
   "reply_end": 15.972,
   "latency_s": 1.347
  },
  {
   "turn": 1,
   "user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
   "user_ref": "<eng>What should I pack for the weather there?</eng>",
   "user_start": 16.672,
   "user_onset": 16.918,
   "backchannels": [],
   "asr": "what should i pack for the weather there",
   "asr_s": 0.055,
   "eot_wait_s": 0.576,
   "speech_end": 18.592,
   "target": {
    "mix": {
     "eng": 0.88,
     "hau": 0.12
    },
    "matrix": "eng",
    "pattern": []
   },
   "target_desc": "Match the user's language mix: English about 90%, Hausa about 10% of the words. The main (matrix) language is English.",
   "reply": "<eng>For Kano, pack light clothes, a hat, and sunscreen—weather is hot and sunny. You might also need a light jacket for evening. Any specific event?</eng> <hau>Wakai kai kai?</hau>",
   "reply_mix": {
    "eng": 0.9,
    "hau": 0.1
   },
   "tvd": 0.016,
   "how": "first_round",
   "first_cand_ok": true,
   "rounds": [
    {
     "round": 0,
     "llm_s": 0.321,
     "cands": [
      {
       "text": "<eng>For Kano, pack light clothes, a hat, and sunscreen—weather is hot and dry. You might also need a scarf for the dust.</eng> <hau>Na gida ne kai?</hau>",
       "ok": true,
       "why": [],
       "tvd": 0.032
      },
      {
       "text": "<eng>For Kano, pack light clothes, a hat, and sunscreen—weather is hot and sunny. You might also need a light jacket for evening. Any specific event?</eng> <hau>Wakai kai kai?</hau>",
       "ok": true,
       "why": [],
       "tvd": 0.016
      },
      {
       "text": "<eng>For Kano tomorrow, pack light clothes—hot and sunny. Bring a hat and sunscreen too.</eng> <hau>Wani kai da kai?</hau>",
       "ok": true,
       "why": [],
       "tvd": 0.096
      }
     ]
    }
   ],
   "mix_s": 0.0055,
   "llm_s": 0.321,
   "tts_first_s": 0.743,
   "spec_start_s": 0.256,
   "compute_to_first_audio_s": 1.142,
   "tts_breakdown": {
    "fe": 0.0,
    "lm": 0.65,
    "flow_voc": 0.22,
    "n_tok": 41,
    "audio_s": 1.64
   },
   "tts_chunks_s": [
    0.743,
    2.614,
    1.588,
    0.958,
    0.871
   ],
   "first_audio": 19.99,
   "reply_end": 34.964,
   "latency_s": 1.398
  },
  {
   "turn": 2,
   "user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]",
   "user_ref": "<hau>To, na gode.</hau>",
   "user_start": 35.664,
   "user_onset": 35.732,
   "backchannels": [],
   "asr": "2 na gode",
   "asr_s": 0.058,
   "eot_wait_s": 0.576,
   "speech_end": 36.864,
   "target": {
    "mix": {
     "hau": 0.76,
     "eng": 0.24
    },
    "matrix": "hau",
    "pattern": [
     "closing:hau"
    ]
   },
   "target_desc": "Match the user's language mix: Hausa about 75%, English about 25% of the words. The main (matrix) language is Hausa. The user closed in Hausa: you may close in Hausa.",
   "reply": "<hau>Sannu!</hau> <eng>Sorry, I did not catch all of that. Could you say it again, or in English?</eng>",
   "reply_mix": {
    "eng": 0.91,
    "hau": 0.09
   },
   "tvd": 0.671,
   "how": "safe_phrasebook",
   "first_cand_ok": false,
   "rounds": [
    {
     "round": 0,
     "llm_s": 0.253,
     "cands": [
      {
       "text": "<hau>Na gode! Wakai kai kai da kai.</hau>",
       "ok": false,
       "why": [
        "missing eng"
       ],
       "tvd": 0.24
      },
      {
       "text": "<hau>Na gode! Kuma kai kai da sani a cikin kai da kai. Kuma ya da ɗan kai kai.</hau>",
       "ok": false,
       "why": [
        "missing eng"
       ],
       "tvd": 0.24
      },
      {
       "text": "<hau>Na gode! Wannan yin kai kai da kai.</hau>",
       "ok": false,
       "why": [
        "missing eng"
       ],
       "tvd": 0.24
      }
     ]
    },
    {
     "round": 1,
     "llm_s": 0.23,
     "cands": [
      {
       "text": "<hau>Na gode!</hau> <eng>My pleasure. Take care while you travel.</eng>",
       "ok": false,
       "why": [
        "mix drift tvd=0.54"
       ],
       "tvd": 0.539
      },
      {
       "text": "<hau>Na gode!</hau> <eng>My pleasure. If you need more tips, just ask.</eng>",
       "ok": false,
       "why": [
        "mix drift tvd=0.58"
       ],
       "tvd": 0.579
      },
      {
       "text": "<hau>Na gode!</hau> <eng>My pleasure. Take care while you travel.</eng> <hau>Wakai kai kai.</hau>",
       "ok": false,
       "why": [
        "mix drift tvd=0.34"
       ],
       "tvd": 0.342
      }
     ]
    }
   ],
   "mix_s": 0.0015,
   "llm_s": 0.483,
   "tts_first_s": 0.64,
   "spec_start_s": 0.256,
   "compute_to_first_audio_s": 1.206,
   "tts_breakdown": {
    "fe": 0.0,
    "lm": 1.42,
    "flow_voc": 0.222,
    "n_tok": 86,
    "audio_s": 3.44
   },
   "tts_chunks_s": [
    0.64,
    1.267,
    1.642
   ],
   "first_audio": 38.325,
   "reply_end": 45.593,
   "latency_s": 1.461
  }
 ]
}