Support more Deepgram settings (diarization?)

[anaxonda]

It would be useful to support more Deepgram transcription features. Being able to toggle diarization would be most useful to me, but also various formatting options the API offers: utterances, punctuation, paragraphs, etc.

Currently if I use the hyperaudio editor with Deepgram I get diarization if I use the 'Enhanced' model, but not the 'Nova-2' or Whisper options.

[maboa]

Great idea! We're also thinking of integrating with Whisper services hosted on fireworks.ai and/or replicate.com. I think we can also build an integration with AssemblyAI.