Stereo Voice File Diarization about whisper-diarization

[KevinGeLe]

I might look into it and make a Pull request. I'm thinking of atlease adding ffmpeg, so it automatically converts to the correct format and adding a parser, so it will connect channel 2 with channel 1 or split them into their own output.

if you want to edit the code now, you just have to add a way to convert the audio file between the input parser and vocal_target.

I think thats what you want?
@Talhazeb

[MahmoudAshraf97]

Hi @Talhazeb , do you mean that each speaker is on a separate channel? if so I don't think diarization is needed since speaker separation is already done

@KevinGeLe feel free to contribute to the repo if you see any changes that need to be done

[Talhazeb]

@KevinGeLe Thanks for informing
@MahmoudAshraf97 How should I proceed with it then. I am actually working on some files which some of them are mono and some of them are stereo. For mono the shared repo works flawless. For stereo, what approach should I take if I have to process it from the same process but have it diarized meaning kind of sequential pyannote rttm file having transcription. If I wouldn't be passsing it through diarization, what approach should I take?

[MahmoudAshraf97]

you can pass both mono and stereo files with no problems, if you need the rttm files, comment the last line in the code that deletes the temp_outputs folder and you'll find it inside nemo/pred_rttms