Add Listen section: 5 playable stereo demos (user left / agent right), refreshed sample logs
2ea57e9 verified | { | |
| "name": "01_igbo_greetings_english", | |
| "desc": "Mostly English with Igbo greetings/closing. Includes a long turn (backchannel) and a barge-in. User turns are TTS-synthesized (Flash, female Igbo voice).", | |
| "user_turns": [ | |
| { | |
| "label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]", | |
| "text": "<ibo>Ndewo!</ibo> <eng>Good morning. Please, can you help me plan my day? I have a lot to do.</eng>", | |
| "wav": "/scratch/duplex_v2/user_audio/01_igbo_greetings_english_turn0.wav", | |
| "orig_label": "TTS-synthesized user turn" | |
| }, | |
| { | |
| "label": "TTS-synthesized user turn (long, mid-turn pause) [identical input audio recovered from the 2026-10-24 demo]", | |
| "text_parts": [ | |
| "<eng>I need to go to the bank, and then the market,</eng>", | |
| "<eng>and in the evening I also want to call my mother.</eng>" | |
| ], | |
| "inner_pause": 0.4, | |
| "wav": "/scratch/duplex_v2/user_audio/01_igbo_greetings_english_turn1.wav", | |
| "text": "<eng>I need to go to the bank, and then the market,</eng> <eng>and in the evening I also want to call my mother.</eng>", | |
| "orig_label": "TTS-synthesized user turn (long, mid-turn pause)" | |
| }, | |
| { | |
| "label": "TTS-synthesized user turn (barge-in 1.5 s into the reply) [identical input audio recovered from the 2026-10-24 demo]", | |
| "text": "<eng>Wait, sorry, the bank closes early today.</eng>", | |
| "barge_at": 1.5, | |
| "wav": "/scratch/duplex_v2/user_audio/01_igbo_greetings_english_turn2.wav", | |
| "orig_label": "TTS-synthesized user turn (barge-in 1.5 s into the reply)" | |
| }, | |
| { | |
| "label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]", | |
| "text": "<eng>Okay, thank you so much.</eng> <ibo>Daalụ!</ibo>", | |
| "wav": "/scratch/duplex_v2/user_audio/01_igbo_greetings_english_turn3.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": "<ibo>Ndewo!</ibo> <eng>Good morning. Please, can you help me plan my day? I have a lot to do.</eng>", | |
| "user_start": 0.7, | |
| "user_onset": 0.762, | |
| "backchannels": [], | |
| "asr": "nde wo good morning please can you help me plan my day i have lot to do", | |
| "asr_s": 0.093, | |
| "eot_wait_s": 0.576, | |
| "speech_end": 5.984, | |
| "target": { | |
| "mix": { | |
| "eng": 0.88, | |
| "ibo": 0.12 | |
| }, | |
| "matrix": "eng", | |
| "pattern": [ | |
| "greeting:ibo", | |
| "inter-sentential" | |
| ] | |
| }, | |
| "target_desc": "Match the user's language mix: English about 90%, Igbo about 10% of the words. The main (matrix) language is English. The user greeted in Igbo: open with a short Igbo greeting.", | |
| "reply": "<ibo>Ụtụtụ ọma!</ibo> <eng>Of course! Let’s start with your most important tasks first. What’s the top thing you need to do today?</eng>", | |
| "reply_mix": { | |
| "eng": 0.86, | |
| "ibo": 0.14 | |
| }, | |
| "tvd": 0.023, | |
| "how": "first_round", | |
| "first_cand_ok": true, | |
| "rounds": [ | |
| { | |
| "round": 0, | |
| "llm_s": 1.106, | |
| "cands": [ | |
| { | |
| "text": "<ibo>Ụtụtụ ọma!</ibo> <eng>Of course! Let’s start with your most important tasks first. What’s the top thing you need to do today?</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.023 | |
| }, | |
| { | |
| "text": "<ibo>Ndewo!</ibo> <eng>Of course! Let’s start with your most urgent tasks. What’s the first thing you need to do today?</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.069 | |
| }, | |
| { | |
| "text": "<ibo>Ndewo!</ibo> <eng>Of course! Let’s start with your most important tasks first. What’s the top thing you need to do today?</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.07 | |
| } | |
| ] | |
| } | |
| ], | |
| "mix_s": 0.0118, | |
| "llm_s": 1.106, | |
| "tts_first_s": 1.196, | |
| "spec_start_s": 0.256, | |
| "compute_to_first_audio_s": 2.42, | |
| "tts_breakdown": { | |
| "fe": 0.0, | |
| "lm": 1.121, | |
| "flow_voc": 0.221, | |
| "n_tok": 66, | |
| "audio_s": 2.64 | |
| }, | |
| "tts_chunks_s": [ | |
| 1.196, | |
| 0.707, | |
| 3.263, | |
| 1.342 | |
| ], | |
| "first_audio": 8.661, | |
| "reply_end": 18.27, | |
| "latency_s": 2.677 | |
| }, | |
| { | |
| "turn": 1, | |
| "user_label": "TTS-synthesized user turn (long, mid-turn pause) [identical input audio recovered from the 2026-10-24 demo]", | |
| "user_ref": "<eng>I need to go to the bank, and then the market,</eng> <eng>and in the evening I also want to call my mother.</eng>", | |
| "user_start": 18.97, | |
| "user_onset": 19.026, | |
| "backchannels": [ | |
| { | |
| "t": 25.28, | |
| "turn_s": 5.28 | |
| } | |
| ], | |
| "backchannel_at": 25.28, | |
| "asr": "i need to go to the bank and then the market and in the evening i also want o call my mother", | |
| "asr_s": 0.093, | |
| "eot_wait_s": 0.576, | |
| "speech_end": 24.992, | |
| "target": { | |
| "mix": { | |
| "eng": 0.96, | |
| "ibo": 0.04 | |
| }, | |
| "matrix": "eng", | |
| "pattern": [] | |
| }, | |
| "target_desc": "Match the user's language mix: English about 95%, Igbo about 5% of the words. The main (matrix) language is English.", | |
| "reply": "<eng>Okay, let’s plan that. Go to the bank early, then the market, and call your mother before dinner. Need help with timings?</eng>", | |
| "reply_mix": { | |
| "eng": 1.0 | |
| }, | |
| "tvd": 0.04, | |
| "how": "first_round", | |
| "first_cand_ok": true, | |
| "rounds": [ | |
| { | |
| "round": 0, | |
| "llm_s": 0.293, | |
| "cands": [ | |
| { | |
| "text": "<eng>Okay, let’s plan that. Go to the bank early, then the market, and call your mother before dinner. Need help with timings?</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.04 | |
| }, | |
| { | |
| "text": "<eng>Okay, let’s plan that. Go to the bank early, then the market, and call your mum after dinner. Should I set a reminder?</eng> <ibo>Ọ dị mma</ibo>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.078 | |
| }, | |
| { | |
| "text": "<eng>Got it. Go to the bank early, then the market, and call your mother after dinner. Should I set a reminder?</eng> <ibo>Ọ dị mma</ibo>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.083 | |
| } | |
| ] | |
| } | |
| ], | |
| "mix_s": 0.0157, | |
| "llm_s": 0.293, | |
| "tts_first_s": 0.963, | |
| "spec_start_s": 0.256, | |
| "compute_to_first_audio_s": 1.383, | |
| "tts_breakdown": { | |
| "fe": 0.0, | |
| "lm": 0.801, | |
| "flow_voc": 0.223, | |
| "n_tok": 50, | |
| "audio_s": 2.0 | |
| }, | |
| "tts_chunks_s": [ | |
| 0.963, | |
| 4.284, | |
| 1.024 | |
| ], | |
| "first_audio": 26.631, | |
| "reply_end": 37.835, | |
| "latency_s": 1.639 | |
| }, | |
| { | |
| "turn": 2, | |
| "user_label": "TTS-synthesized user turn (barge-in 1.5 s into the reply) [identical input audio recovered from the 2026-10-24 demo]", | |
| "user_ref": "<eng>Wait, sorry, the bank closes early today.</eng>", | |
| "user_start": 28.131, | |
| "user_onset": 28.2, | |
| "backchannels": [], | |
| "barge": { | |
| "user_onset": 28.2, | |
| "detected": 28.384, | |
| "agent_stopped": 28.424, | |
| "reaction_s": 0.224, | |
| "agent_chunks_played": 1, | |
| "agent_chunks_total": 3 | |
| }, | |
| "asr": "wait sorry the band closes early today", | |
| "asr_s": 0.074, | |
| "eot_wait_s": 0.576, | |
| "speech_end": 30.528, | |
| "target": { | |
| "mix": { | |
| "eng": 0.98 | |
| }, | |
| "matrix": "eng", | |
| "pattern": [] | |
| }, | |
| "target_desc": "Match the user's language mix: English about 100% of the words. The main (matrix) language is English. Reply only in English.", | |
| "reply": "<eng>Okay, no problem. Let’s adjust—go to the market first, then call your mum before the bank closes.</eng>", | |
| "reply_mix": { | |
| "eng": 1.0 | |
| }, | |
| "tvd": 0.01, | |
| "how": "first_round", | |
| "first_cand_ok": true, | |
| "rounds": [ | |
| { | |
| "round": 0, | |
| "llm_s": 0.208, | |
| "cands": [ | |
| { | |
| "text": "<eng>Okay, no problem. Let’s adjust—go to the market first, then call your mum before the bank closes.</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.01 | |
| }, | |
| { | |
| "text": "<eng>Okay, no problem. Let’s shift the bank visit to morning, then market, and call your mum after dinner.</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.01 | |
| }, | |
| { | |
| "text": "<eng>Okay, no problem. Let’s shift the bank trip to earlier today. What time can you make it before it closes?</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.01 | |
| } | |
| ] | |
| } | |
| ], | |
| "mix_s": 0.0049, | |
| "llm_s": 0.208, | |
| "tts_first_s": 0.908, | |
| "spec_start_s": 0.256, | |
| "compute_to_first_audio_s": 1.209, | |
| "tts_breakdown": { | |
| "fe": 0.0, | |
| "lm": 2.513, | |
| "flow_voc": 0.285, | |
| "n_tok": 130, | |
| "audio_s": 5.2 | |
| }, | |
| "tts_chunks_s": [ | |
| 0.908, | |
| 2.798 | |
| ], | |
| "first_audio": 31.993, | |
| "reply_end": 39.991, | |
| "latency_s": 1.465 | |
| }, | |
| { | |
| "turn": 3, | |
| "user_label": "TTS-synthesized user turn [identical input audio recovered from the 2026-10-24 demo]", | |
| "user_ref": "<eng>Okay, thank you so much.</eng> <ibo>Daalụ!</ibo>", | |
| "user_start": 40.691, | |
| "user_onset": 40.753, | |
| "backchannels": [], | |
| "asr": "ok thank you so much talo", | |
| "asr_s": 0.05, | |
| "eot_wait_s": 0.576, | |
| "speech_end": 43.104, | |
| "target": { | |
| "mix": { | |
| "eng": 0.99 | |
| }, | |
| "matrix": "eng", | |
| "pattern": [] | |
| }, | |
| "target_desc": "Match the user's language mix: English about 100% of the words. The main (matrix) language is English. Reply only in English.", | |
| "reply": "<eng>You're very welcome, Talos!</eng>", | |
| "reply_mix": { | |
| "eng": 1.0 | |
| }, | |
| "tvd": 0.005, | |
| "how": "first_round", | |
| "first_cand_ok": true, | |
| "rounds": [ | |
| { | |
| "round": 0, | |
| "llm_s": 0.17, | |
| "cands": [ | |
| { | |
| "text": "<eng>You're very welcome, Talos!</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.005 | |
| }, | |
| { | |
| "text": "<eng>You’re welcome, Talu! Have a great day—stay calm and organized.</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.005 | |
| }, | |
| { | |
| "text": "<eng>You’re welcome, Tallo! Have a great day.</eng>", | |
| "ok": true, | |
| "why": [], | |
| "tvd": 0.005 | |
| } | |
| ] | |
| } | |
| ], | |
| "mix_s": 0.0042, | |
| "llm_s": 0.17, | |
| "tts_first_s": 1.361, | |
| "spec_start_s": 0.256, | |
| "compute_to_first_audio_s": 1.593, | |
| "tts_breakdown": { | |
| "fe": 0.0, | |
| "lm": 1.138, | |
| "flow_voc": 0.223, | |
| "n_tok": 67, | |
| "audio_s": 2.68 | |
| }, | |
| "tts_chunks_s": [ | |
| 1.361 | |
| ], | |
| "first_audio": 44.953, | |
| "reply_end": 47.633, | |
| "latency_s": 1.849 | |
| } | |
| ] | |
| } |