WAV file
Running model inference...
Pipeline
Multi-boundary detection with peek-ahead policy.
Verifier: TitaNet local sim_boundary < 0.40.
Speaker re-id by segment-level clustering.
Waveform
SC head probability stream (live)
Select a file and click Run
Speaker timeline
| # | Time | Peak | sim_boundary | vs GT |
|---|
ASR Transcript colored by detected speaker, live as audio plays
Select a file and run inference to see transcript...