search

nkilm / offline-whisperx

Run different pipelines of WhisperX - Transcription, Diarization, VAD, Alignment completely OFFLINE.
24 stars 1 forks source link

Working together on whisper GUI #5

Open chaoliu-cl opened 1 month ago

chaoliu-cl commented 1 month ago

Hi @nkilm,

Thank you for your excellent work on the offline WhisperX pipeline! Your repository has been a great inspiration for me, and I’m currently developing a Python GUI (https://github.com/chaoliu-cl/whisper_transcriber) to simplify the use of WhisperX. However, I've encountered an issue with Speaker Diarization when trying to run it offline. Although I’ve stored the models in the correct directory, I’m unable to get it to work.

Would you mind taking a look and offering some guidance?

Best regards, Chao

nkilm commented 1 month ago

Hi @chaoliu-cl! Thank you. I'm glad that you found the script to be useful!

https://github.com/m-bain/whisperx does not have the support for running the audio pipelines in offline mode(without internet connection). That's why I had to make this wrapper script.

However, I've encountered an issue with Speaker Diarization when trying to run it offline.

You can try using the https://github.com/nkilm/offline-whisperx/blob/main/pyannote_diarization.py class with support for offline diarization. Also please make sure that you are using whisperx >= 3.1.3.

I hope this helps. Regards, Nikhil