Feature Request: Possibility to add Piper and Piper Audiobook export

[Aamir3d]

Hello,
I saw this TTS engine called Piper here https://github.com/rhasspy/piper

Is it possible to add Piper and this additional https://github.com/thillsd/piperbook (Epub to MP3) to the TTS Generation WebUI?
I tried installing this separately, but it's not easy to work with.
Alternatively, this is another option I saw https://github.com/aedocw/epub2tts

The purpose is to convert some ebooks I have into audio books with better quality.

[rsxdalv]

This looks great! MIT license is compatible and hopefully the project as well.
I am a bit busy until the weekend, but I hope I can soon get back to making stuff for the repo.

[Aamir3d]

Thanks @rsxdalv , appreciate this!

[Aamir3d]

Here's another project https://github.com/aedocw/epub2tts
This is based off https://github.com/coqui-ai/TTS
The assumption is that we can get long form TTS working with different voices.

[rsxdalv]

Here's another project https://github.com/aedocw/epub2tts This is based off https://github.com/coqui-ai/TTS The assumption is that we can get long form TTS working with different voices.

Coqui has a copyleft license, and is thus incompatible with this project. The epub2tts project might be usable without risking the MPL but that's something to be aware of. Edit: To be clear, using the epub2tts project with coqui (as it is by default) is basically a no go.

[Aamir3d]

Thanks for the explanation @rsxdalv !! Looks like the Piper model might work with the GUI. Thinking beyond only epubs, is there (possibly) a way to export PDF, WORD, RTF and TXT with the TTS GUI?

As always, appreciate all your hard work on this project. (Do you have a LinkedIn - I would like to make a post sometime attributing this GUI to you and showcasing what it does).

[rsxdalv]

Piper with it's MIT license looks good.

Do you mean PDF to mp3 kind of an approach?

Here's my linkedin, but as you can see it's basically abandoned: https://www.linkedin.com/in/roberts-slisans-1829aba3

[Maki9009]

has this been implemented you to convert a pdf to an audio book?

[rsxdalv]

has this been implemented you to convert a pdf to an audio book?

PDF needs to be converted first. Some PDFs might not convert to a proper text (like word file) nicely. You should convert it to epub and then see if the https://github.com/aedocw/epub2tts project works for you.
By the way, you are using Tortoise for this, right?

[terion-name]

any progress on adding piper?(( it sounds better then coqui and coqui has been closed

[DrewThomasson]

Just made a Audiobook Generator that uses Piper-tts,

You can select from all the Available Piper tts voices in the gui.

https://github.com/DrewThomasson/ebook2audiobookpiper-tts

https://hub.docker.com/repository/docker/athomasson2/ebook2audiobookpiper-tts

Has its own docker image and everything lol

Idk if y'all would find that helpful but anyway, there you go,

It uses Calibre to convert any document you give it to txt so PDF, epub, Word etc should work with it

[rsxdalv]

Ok, I tried to install the project but it's fairly broken. Until piper devs address rhasspy/piper#509 it will just be a pain. I managed to build the C++ library; however the instructions claimed that it would generate a python wheel which does not happen. So as it stands, there seems to be no clear instruction about how to install piper tts on Python 3.10.11 or almost any Python version for that matter unless it matches exactly what the devs used.

Building from source could work in theory, but unfortunately in practice there's a "gap" between piper-phonemize the C++ project and piper-phonemize the pip package.