search

thewh1teagle / vibe

Transcribe on your own!
https://thewh1teagle.github.io/vibe/
MIT License
1.25k stars 73 forks source link

[Feature Request]: Ffmpeg - audio filter option #362

Closed easyfab closed 1 week ago

easyfab commented 2 weeks ago

Describe the feature

ffmpeg process is really to slow with -af loudnorm=I=-16:TP=-1.5:LRA=11 (If I read correclty audio.rs )
More than 8 minutes for a 240 min audio files

It take longer for the ffmpeg process than the whisper process for me !

Is there a way to disable this filter or to change this to a more faster filter like -af dynaudnorm ?

with -af dynaudnorm it's a ~500x speed vs ~25x speed for -af loudnorm=I=-16:TP=-1.5:LRA=11

thewh1teagle commented 1 week ago

I may add this option in the furue, meanwhile you can normalize manually the file before transcribing and the app will skip it

easyfab commented 1 week ago

Oh, I didn't know that using a wav file you can bypass the ffmpeg process. I tried first without -ac 1 and it didn't work but with -ac 1 it's ok. the option is not necessary in this case.

thewh1teagle commented 1 week ago

Added in https://github.com/thewh1teagle/vibe/releases/tag/v2.6.7 including option for custom ffmpeg command

easyfab commented 1 week ago

Nice. Thank you.