Works well for songs, but it kinda messes up Audiobooks

[kainbock1]

Or am I missing good settings here?

[JackismyShephard]

If the audiobook contains just one speaker and no other sources of sound, then you don't really need the full pipeline that this project implements. Specifically, this implementation does a lot of pre-processing to separate vocals in the given input. Maybe this is what is causing problems for you. I am considering making some modifications to the code that will make it work well for other applications such as audiobook conversion and TTS.

[kainbock1]

Would love that :)