Concept

Speaker diarization

Enable Pro speaker identification and consume optional segment speaker identifiers.

Enable diarization

Request field
{
  "source_url": "https://cdn.example.com/panel.mp3",
  "speaker_diarization": true
}

The flag defaults to false and requires a Pro account. Unsupported accounts receive 403 pro_required before work is queued.

Output

Segment with speaker
{
  "start": 18.4,
  "end": 23.1,
  "text": "Let's move to the next question.",
  "speaker": 1
}

Speaker identifiers are associated with timestamped segments and may be numbers or strings. They are not guaranteed to be display-ready labels. The top-level full transcript remains a plain text string.

Client handling

  • Treat speaker as optional on each segment.
  • Do not assume a fixed number or naming convention for speakers.
  • Preserve timestamps when grouping adjacent segments by speaker.
© 2026 FastTranscriberBuilt for developers who work with speech.