Skip to main content

Speaker diarization

Speaker diarization separates who spoke when in Batch STT. Enable by setting use_diarization: true in the request. If not enabled, diarization options are ignored. If you know the number of speakers, set diarization.spk_count (e.g., 2 for phone calls). If omitted, the service will estimate.

caution

When multi-channel is used, diarization is disabled. The spk will follow channel ids.

Usage​

In the POST Request body, set use_diarization: true. If needed, configure diarization fields such as diarization.spk_count (e.g., 2).

POST example: Example

Unified example (presets): Example