original: https://github.com/shivammehta25/Matcha-TTS
huggingface demo: https://huggingface.co/spaces/doof-ferb/MatchaTTS_ngngngan
only for inference, do not use to train
i don’t use ONNX because it cannot change number of ODE steps
git clone then install: pip install -e . --find-links=https://download.pytorch.org/whl/torch_stable.html
download vocoder:
hifigan_univ_v1(recommended): https://github.com/shivammehta25/Matcha-TTS-checkpoints/releases/download/v1.0/g_02500000- or
hifigan_T2_v1: https://github.com/shivammehta25/Matcha-TTS-checkpoints/releases/download/v1.0/generator_v1
download checkpoints: https://huggingface.co/doof-ferb/matcha_ngngngan
CLI usage: python matcha/cli.py --vocoder_path=… --checkpoint_path=… --output_folder=outputs --text=…
GUI usage: python matcha/app.py
