Labmem-Zhouyx / PyTorch_character_CN_Taco2

A PyTorch inplementation of character-based Tacotron2 for Chinese/Mandarin

character-based Tacotron2 for Chinese

An implementation of Tacotron2 based Pytorch

Input: Chinese Character Sequence

Output: Mel-spectrogram

1. Prepare training data: text and corresponding mel-spectrogram files.

Referred：https://github.com/Labmem-Zhouyx/audio2mel_preprocessor

2. Train the model.

python train.py -o save -l logs

3. Inference.

python inference.py -t 'sentences.txt' -c ./save/checkpoint_100000

Reference

https://github.com/thuhcsi/tacotron