bark icon indicating copy to clipboard operation
bark copied to clipboard

:pinching_hand: Better silence management/encoding

Open adriens opened this issue 1 year ago • 3 comments

... so output wav can be easily previewed on various platforms and easily compressed to mp3/webm :

  • https://github.com/suno-ai/bark/issues/288

adriens avatar May 17 '23 22:05 adriens

not sure i understand the title of this PR..

gkucsko avatar May 18 '23 13:05 gkucsko

In fact, this PR is to make a better slience management so output wav can be easily compressed to mp3 or webm

adriens avatar May 18 '23 21:05 adriens

I think I can translate. He's probably getting 32 or 64 bit wav files however he's running or integrating that code, because it's matching the type from np.zeros silence segments. The files are bigger and some players don't like them, windows media player won't play the 32 bit ones. Kind of a implementation detail though, not really a problem.

JonathanFly avatar May 23 '23 09:05 JonathanFly

Screenshot 2023-06-23 at 12 07 30 PM

This PR should be good to land IMO.

tongbaojia avatar Jun 23 '23 16:06 tongbaojia