search

p0n1 / epub_to_audiobook

EPUB to audiobook converter, optimized for Audiobookshelf
MIT License
1.15k stars 118 forks source link

Piper TTS support #77

Closed vcalv closed 3 months ago

vcalv commented 4 months ago

piper TTS is an excellent TTS system that you can run on your own machine capable of creating very natural speech in real-time using only CPU and no GPU acceleration even in low end machines.

This means that one does not need to upload anything to any cloud service.

There are many models for many languages and some have many hundreds of different voices.

This PR adds support for this back-end and some documentation.

It's fully functional but I'm sure there's room for improvement.

Thank you.

Bryksin commented 4 months ago

Heh I didn't monitor the project for some time since major init refactoring As I see we have many opened PRs and tons of issues but @p0n1 seems like not maintaining much, but I don't have permission to accept PRs

p0n1 commented 4 months ago

Thanks for contributing @vcalv. There was a piper feature contribution before our code refactoring, so we can't merge that one directly. I'm happy to test and merge your contributions when I'm available.

p0n1 commented 4 months ago

Heh I didn't monitor the project for some time since major init refactoring As I see we have many opened PRs and tons of issues but @p0n1 seems like not maintaining much, but I don't have permission to accept PRs

Hey @Bryksin. It has been a while. I'm still here. Yeah, tons of pending work wait for us. I'd be happy if you could help.

vcalv commented 4 months ago

Thanks for contributing @vcalv. There was a piper feature contribution before our code refactoring, so we can't merge that one directly. I'm happy to test and merge your contributions when I'm available.

Hi.

Just to make things clear.

This PR is exclusively for piper tts and maps all parameters that can be mapped for that provider. Things like pause time, speed, etc.

The other one is #18 which is a generic local tts solution that could also work with piper but can use any command that outputs sound files from text.

These are not the same thing and with small changes to #18 (not defaulting to piper) both could and should be merged.

Bryksin commented 3 months ago

in relation to #18 - it is super outdated and implemented for project before major refactoring, it cannot be accepted