phoonnx-mixertts / README.md
Jarbas's picture
docs: model card
5aff21b verified
|
Raw
History Blame Contribute Delete
652 Bytes
metadata
license: other
pipeline_tag: text-to-speech
library_name: phoonnx
tags:
  - phoonnx
  - mixer-tts
  - tts

phoonnx-mixertts

Mixer-TTS voices (NVIDIA MLP-Mixer/FastPitch acoustic, text->80-ch mel) for phoonnx. Two-stage; voiced via a vocoder (Griffin-Lim by default). One voice per subfolder (model.onnx + native config.json + vocoder.json). Mirror of the nipponjo/mixer-tts-pytorch LJSpeech checkpoints (en, 22kHz). Part of the phoonnx community mirror.