faster-whisper implementation

[jooni22]

Hi, you should think about faster-whisper implementation. IMHO the best way in speech to transcribe method right now.
https://github.com/guillaumekln/faster-whisper

[absadiki]

Hi @jooni22,

Thanks for bringing it to my attention. faster-whisper seems really good.
I will add it as soon as possible.

Thanks!

[absadiki]

Hi @jooni22,

Support for faster-whisper has been added.
Let me know if you find any issues.

Thanks again!

[shubhendumadhukar]

Anyway I can enable logging for faster-whisper? While using whisper or whispercpp, I can see the progress/percentage. But not the case with faster-whisper? No error, so I am not sure if it's stuck. I am new to this, not really sure if a progress bar is supported in faster-whisper implementation.

[absadiki]

@shubhendumadhukar, are you using faster-whisper from Python or the web-ui ?

[shubhendumadhukar]

I'm using it from the web ui.

[absadiki]

@shubhendumadhukar I don't think faster-whisper provides a helpful progress as other models. But I have activated debugging level. It shows the time of the segment being processed. Update subsai to the newest commit and give it a try.

[jooni22]

Very fast implementation I will test the new model then let you know.