Speaker Labels

Identify and label different speakers in multi-voice recordings with Whisper, Deepgram, or Mistral.

For conversations with more than one voice, VivaDicta can identify and label different speakers in the transcript. You get a speaker-separated note instead of one long wall of text, which makes meetings, interviews, and calls much easier to read afterwards.

Supported Models

Speaker labels work with local Whisper and the cloud providers Deepgram and Mistral. When any of these is your active transcription model, turning the setting on adds speaker tags to the output. Other models ignore the setting.

Turn It On

Open Settings and scroll to the Transcription section. Toggle Speaker Labels on. The change applies to new recordings - existing notes are not re-processed.

Settings screen with Speaker Labels toggle under Transcription

Tips for Better Results

Record with a microphone close to each speaker when possible - clean audio gives the best diarization.
Longer recordings with distinct voices separate better than short clips with overlapping speech.
If you only ever dictate solo notes, keep this off to save a little processing time.

See also: Transcription Models for picking a compatible provider.