Mic audio streams in real-time via WebSocket to the Whisper transcription engine.
Interim results appear in gray while speech is still being analyzed.
Final results appear in white once a segment is stable.
Live sessions are not saved to history. Use Upload for saved transcriptions.
WebSocket endpoint:
ws://host/v1/listen?token=…&model=base&encoding=webm
Audio is converted to 16 kHz mono PCM before transcription.
Accepted: WAV, MP3, OGG, FLAC, WebM, M4A, MP4, OPUS
Max size: 50 MB
Results are saved to History and auto-deleted after the configured retention period.
REST endpoint:
POST /v1/listen?model=base&language=auto