- Fix voice_profiler threshold bug (HOST label overwrote Unknown unconditionally) - Audio preload optimization: single ffmpeg per episode, 149.5x realtime on 5070 Ti - WavLM threshold raised to 0.85 (Mike 0.90-0.99, callers 0.46-0.83) - Promo/bumper filter: weighted signature scoring, 42->27 clean Q&A pairs - Text-only Q&A fallback for episodes with no CALLER diarization labels - TRANSFORMERS_OFFLINE=1 to skip HuggingFace freshness checks - Add diarize_2018.py for targeted re-run + FTS5 rebuild - Add benchmark.py + BENCH_SETUP.md for GURU-BEAST-ROG (RTX 4090) comparison - Commit 9-episode training diarization.json outputs - Session log: 2026-04-27-diarization-pipeline.md Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
327 lines
6.0 KiB
JSON
327 lines
6.0 KiB
JSON
{
|
|
"num_speakers": 2,
|
|
"speaker_map": {
|
|
"HOST": "HOST",
|
|
"CALLER": "CALLER"
|
|
},
|
|
"turns": [
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 0.0,
|
|
"end": 30.0,
|
|
"confidence": 0.87
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 25.0,
|
|
"end": 35.0,
|
|
"confidence": 0.79
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 30.0,
|
|
"end": 105.0,
|
|
"confidence": 0.95
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 100.0,
|
|
"end": 115.0,
|
|
"confidence": 0.56
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 110.0,
|
|
"end": 160.0,
|
|
"confidence": 0.96
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 155.0,
|
|
"end": 170.0,
|
|
"confidence": 0.64
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 165.0,
|
|
"end": 210.0,
|
|
"confidence": 0.9
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 205.0,
|
|
"end": 225.0,
|
|
"confidence": 0.79
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 220.0,
|
|
"end": 235.0,
|
|
"confidence": 0.98
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 230.0,
|
|
"end": 240.0,
|
|
"confidence": 0.85
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 235.0,
|
|
"end": 290.0,
|
|
"confidence": 0.97
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 285.0,
|
|
"end": 320.0,
|
|
"confidence": 0.55
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 315.0,
|
|
"end": 415.0,
|
|
"confidence": 0.98
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 410.0,
|
|
"end": 420.0,
|
|
"confidence": 0.8
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 415.0,
|
|
"end": 440.0,
|
|
"confidence": 0.97
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 435.0,
|
|
"end": 455.0,
|
|
"confidence": 0.57
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 450.0,
|
|
"end": 530.0,
|
|
"confidence": 0.96
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 525.0,
|
|
"end": 535.0,
|
|
"confidence": 0.84
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 530.0,
|
|
"end": 595.0,
|
|
"confidence": 0.88
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 590.0,
|
|
"end": 605.0,
|
|
"confidence": 0.84
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 600.0,
|
|
"end": 870.0,
|
|
"confidence": 0.89
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 865.0,
|
|
"end": 930.0,
|
|
"confidence": 0.6
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 925.0,
|
|
"end": 945.0,
|
|
"confidence": 0.87
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 940.0,
|
|
"end": 970.0,
|
|
"confidence": 0.54
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 965.0,
|
|
"end": 1015.0,
|
|
"confidence": 0.95
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1010.0,
|
|
"end": 1040.0,
|
|
"confidence": 0.49
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1035.0,
|
|
"end": 1060.0,
|
|
"confidence": 0.98
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1055.0,
|
|
"end": 1070.0,
|
|
"confidence": 0.65
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1065.0,
|
|
"end": 1105.0,
|
|
"confidence": 0.88
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1100.0,
|
|
"end": 1120.0,
|
|
"confidence": 0.82
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1115.0,
|
|
"end": 1225.0,
|
|
"confidence": 0.96
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1220.0,
|
|
"end": 1230.0,
|
|
"confidence": 0.84
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1225.0,
|
|
"end": 1265.0,
|
|
"confidence": 0.96
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1260.0,
|
|
"end": 1275.0,
|
|
"confidence": 0.62
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1270.0,
|
|
"end": 1310.0,
|
|
"confidence": 0.97
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1305.0,
|
|
"end": 1320.0,
|
|
"confidence": 0.57
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1315.0,
|
|
"end": 1340.0,
|
|
"confidence": 0.94
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1335.0,
|
|
"end": 1350.0,
|
|
"confidence": 0.57
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1345.0,
|
|
"end": 1705.0,
|
|
"confidence": 0.86
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1700.0,
|
|
"end": 1725.0,
|
|
"confidence": 0.65
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1720.0,
|
|
"end": 1825.0,
|
|
"confidence": 0.96
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1820.0,
|
|
"end": 1840.0,
|
|
"confidence": 0.65
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1835.0,
|
|
"end": 1900.0,
|
|
"confidence": 0.92
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1895.0,
|
|
"end": 1925.0,
|
|
"confidence": 0.83
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1920.0,
|
|
"end": 1935.0,
|
|
"confidence": 0.86
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1930.0,
|
|
"end": 1940.0,
|
|
"confidence": 0.85
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1935.0,
|
|
"end": 1955.0,
|
|
"confidence": 0.88
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 1950.0,
|
|
"end": 1960.0,
|
|
"confidence": 0.81
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 1955.0,
|
|
"end": 2050.0,
|
|
"confidence": 0.98
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 2045.0,
|
|
"end": 2055.0,
|
|
"confidence": 0.49
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 2050.0,
|
|
"end": 2455.0,
|
|
"confidence": 0.9
|
|
},
|
|
{
|
|
"speaker": "CALLER",
|
|
"start": 2450.0,
|
|
"end": 2465.0,
|
|
"confidence": 0.83
|
|
},
|
|
{
|
|
"speaker": "HOST",
|
|
"start": 2460.0,
|
|
"end": 2675.0,
|
|
"confidence": 0.87
|
|
}
|
|
]
|
|
} |