← sources

Been working on a universal tts finetuner for a bit so far works for 17 tts engines (xtts,coqui stuff, piper,etc), docker or cli

source: reddit (reddit:selfhosted)published 2026-09-26 05:27ingested 2026-09-27 02:38

retained text

Little side project I’ve been doing outside of [ebook2audiobook](https://github.com/DrewThomasson/ebook2audiobook/)but anyway idk I guess it’s ready to share at this point 

Web GUI or CLI

Has a docker cause yall like that

So far the tts engines I’ve got supported are 

XTTS v2, XTTS v1, Piper TTS, VITS, MMS / Fairseq VITS, FastPitch,            
  FastSpeech 2, Glow-TTS, DelightfulTTS, Tacotron2 DCA, Tacotron2 DDC,         
  Tacotron2 Capacitron, Overflow, SpeedySpeech, FastSpeech, Align TTS,         
  NeuralHMM-TTS

You can even use ebook2audiobook to generate synthetic data of 1 voice cloning to then use to distill the voice in a (ittty bitty) piper model

I’ll be making a video on it soon

Hopefully this isn’t too niche

ANYWAY wanted to see y’all’s thoughts on it enjoyyyyy

document fetched 2026-09-27 00:30 · canonical https://reddit.com/r/selfhosted/comments/1wqi8re/been_working_on_a_universal_tts_finetuner_for_a