UIS-RNN base diarization pipeline

[malarinv]

The current diarization model/pipeline is based on clustering and it seems to have its own problems (All samples have mutually equal similarities. Returning arbitrary cluster center(s) etc.) .

Is there any plan to implement UIS-RNN based (Fully supervised diarization) model pipeline?

[hbredin]

I have no such plan. Why don't you use the linked UIS-RNN repo directly since it has been implemented there already?

[malarinv]

I am using that already, but the api interface/pipeline is much cleaner with pyannote.