feat: add offline diarization support with configuration file

[semidark]

• Introduced OfflineDiarizationPipeline class for local speaker diarization using a specified configuration file.
• Added command-line arguments to enable offline diarization and specify the config file path in transcribe.py.
• Created pyannote_diarization_config.yaml for configuring the diarization pipeline parameters.

[ahmedcareweb]

Any updates on this one? I see there have been activity lately on the repo. I hope this one gets attention

[fcakyon]

@Barabazs can you have a look at this pr? 😊

[ahmedcareweb]

I see that the repo is getting some attention back. 18 commits already in the last month. Can @semidark resolve the conflicts so this pr can be merged?

[semidark]

I can have a look at it next week. I was running my fork for 11 Months straight in my customers air gapped production environment. No problems there.

[LukeP-Flanagan]

This is exactly the fix I am currently looking for. I eagerly look forward to the conflict resolutions so that this can be merged. Or, alternatively, the updated fork being public.
Do you hard-code CPU-only processing? (In whisperx/offline_diarize.py --> class OfflineDiarizationPipeline:
def init(
self,
config_path,
device="cpu",
): )

Looking forward to hearing progress.