Skip to main content

Backends

Two transcription backends are available. Admins can switch between them at runtime via the System page in the web app — no code deployment needed.

Local (Whisper.cpp)

Runs entirely on the worker machine. Audio never leaves your infrastructure.
  • Model files must be present in the transcription_models/ volume
  • The worker must restart to load the model (switching back to local from Scaleway requires a restart)
  • Accuracy depends on the model size loaded

Scaleway (remote inference)

Audio is posted to Scaleway Generative APIs ({base_url}/audio/transcriptions). Results come back in verbose_json format.
  • base_url and api_key are set in config.yml (ai_inference.scaleway) — not editable from the admin UI (security boundary)
  • The realtime model and final model can be set separately from the admin UI
Size limit: single-file session audio uploads are capped at 200 MiB. Long sessions must use chunked upload — a single-file upload of a multi-hour recording will fail.

Speaker labels

Transcription does not diarise. Whisper segments carry no speaker label, so a transcript from a single microphone is unattributed. Labels exist only for sources that supply them: Recall and phone/track names. Speaker claims and relabelling operate on the labels present on segments.

Runtime configuration (admin)

Admins can change the following at runtime from Admin → System → Transcription: base_url and api_key are config-file only — not exposed in the admin UI.

Transcript format

Transcripts are stored as structured segments:
Speaker labels are session-scoped strings. There is no cross-session speaker identity.

Re-running failed transcription

Individual failed chunk jobs can be re-run from Admin → Jobs → Transcriptions. Select the session, expand the chunk group, and click “Re-run” on the failed chunk.