Add Speaker diarisation /speaker detection for interview trascription

[menelic]

Also mentioned in #469 This is implemented in this Whisper gui built in streamlit: https://github.com/jojojaeger/whisper-streamlit (you find the diarisation version here https://github.com/jojojaeger/whisper-streamlit/tree/master/whisper-streamlit-speaker but info on to in readme at the first link) first link) Because yours is a cross platform desktop app, this can become a go-to for many journalists, researchers etc for whom such a feature would be key.

[bfrye26]

I would love this, it is such an easy app to use, and if it had this feature it would be something I use daily!

[johnfelipe]

Pls add this feature

[marrie]

It would actually be wonderful to do this even if it was just "speaker 1" "speaker 2" etc. so
speaker 1: american 1040 requesting IFR
speaker 2: American 1040 go ahead.

you might be able to clean up the transcript then in VsCode for clarity. Thoughts?