Matcha-TTS-Japanese Examples
some demos use onnx and possible error on your platforms
Basic Examples
Match-TTS-ModelInfo - Just print Onnx input and output Info and Local only Onnx
English Examples
phonemizer is different,but the model with official one would works well
Even Using GPU quantized model are 3-5 time slow than huggingface models
Matcha-TTS VCTK 109 voices Speak(GPU) - faster GPU Version
Matcha-TTS VCTK 109 voices Speak(CPU) - try this if gpu error,slow cpu version
Matcha-TTS LJSpeech single voice Speak(GPU) - best single speaker GPU Version
Matcha-TTS En001 single voice Speak(GPU) - My trained model from CommonVoice dataset with ljspeech pretrained
Other related Examples
Convert Cmudict to IPA - create cumdict-ipa_simplified_vowel-aligned.tsv