phoonnx-coqui-vits

phoonnx-coqui-vits

Mirror of the Coqui TTS project's community VITS checkpoints, exported to ONNX for phoonnx, OpenVoiceOS's offline TTS library. These are the single-speaker and multi-speaker VITS models that shipped in Coqui TTS's own model zoo (.models.json), trained on Common Voice, CSS10, M-AILABS, and OpenBible/Bloom-Library corpora, covering 34 languages from Bulgarian to Yoruba.

This repository holds the raw ONNX weights and config.json files consumed by phoonnx's voice_index (coqui_vits.json); it is not meant to be browsed for voices directly — use phoonnx's voice manager, which resolves voice IDs to the files here.

Usage

import wave
from phoonnx.model_manager import TTSModelManager

manager = TTSModelManager()
manager.load()
manager.merge_default_voices()

voice = manager.voices["coqui/de-thorsten-vits"].load()
with wave.open("out.wav", "wb") as wav_file:
    voice.synthesize_wav("Guten Tag, wie geht es dir?", wav_file)

License

license: other because each checkpoint carries the license Coqui TTS assigned it in its own .models.json at the time of training — these differ per corpus and are not uniform. Below is every voice in this repository with the license recorded by upstream Coqui TTS.

Licenses are as recorded in the upstream coqui-ai/TTS .models.json at the time these checkpoints were exported; phoonnx itself is Apache-2.0.