Closed DawgZter closed 8 months ago
Getting diarization from Deepgram is fairly simple like you said, a matter of providing the correct config settings. The code would need to be adjusted to support from current You, Speaker, to You, Speaker-1, Speaker-2, ..... Speaker-N.
I would be happy to help anyone if they want to take this on.
Deepgram has been upgraded to version 3.1.x in case it helps in anyway. Diarization is added to an example
Could be interesting to include diarization support, so the LLM has the context of multiple speakers and which parts are spoken by who. I know deepgram already supports this in their transcription though, so maybe it's just a matter of changing up the settings on the api call and it's good to go? https://developers.deepgram.com/docs/diarization