Does Benecaster identify who’s speaking in transcripts?
Yes — this is called speaker diarization, and Benecaster supports it for multi-speaker shows like interviews and co-hosted podcasts. When enabled, the transcript labels each speaker throughout (“Speaker 1,” “Speaker 2,” etc.) rather than presenting a single unbroken block of text.
You enable diarization per show — it’s off by default, since solo shows don’t need it and it adds a small amount of processing time. An optional speaker count hint (set it to 2 for a standard host + guest format) keeps the labeling clean on long recordings.
Speaker support by provider:
| Provider | Diarization |
|---|---|
| AssemblyAI | ✓ Native |
| Deepgram | ✓ Native |
| Rev.ai | ✓ Native |
| OpenAI (gpt-4o-transcribe) | ✓ Native |
| OpenAI Whisper (original) | ✗ Not supported — upgrade to the gpt-4o-transcribe model to enable it |
Named speakers with Guest Manager: If you have the Guest Manager add-on installed, you can map speaker labels to real names directly in the transcript editor — “Speaker 1” becomes your name and “Speaker 2” becomes your guest’s name. Those names then appear in the displayed transcript, in SRT/VTT exports, and in the <podcast:transcript> RSS tag that Podcasting 2.0 apps display.
