Non-autoregressive Deep Learning-Based TTS Template

This is a template for the Non-autoregressive TTS model. It contains

• Data Preprocessing Pipeline
• Data Loader
• Model / Trainer
• Logger, Postprocessing (logging, synthesizing, plotting, etc..)

How to use it?

1. Clone the repository.

   git clone https://github.com/keonlee9420/Deep-Learning-TTS-Template
   cd Deep-Learning-TTS-Template
   

2. Replace all MYMODEL strings in this repo with your model name and also rename the file model/MYMODEL.py.
3. Build your model on model/ and check train.py and synthesize.py.
4. Use README_template.md for the README.md file of your project.
5. Feel free to add /img for your model architecture and tensorboard examples. It would also be nice to show your model's output audio in /demo.
6. Don't forget to update requirements.txt and /config of your project.

Citation

@misc{lee2021deep_learning_tts_template,
  author = {Lee, Keon},
  title = {Deep-Learning-TTS-Template},
  year = {2021},
  publisher = {GitHub},
  journal = {GitHub repository},
  howpublished = {\url{https://github.com/keonlee9420/Deep-Learning-TTS-Template}}
}

References

• ming024's FastSpeech2