This is a PyTorch implementation of Interspeech 2023 paper Diverse and Expressive Speech Prosody Prediction with Denoising Diffusion Probabilistic Model.

Audio Samples

Audio samples generated by this implementation can be found here.

References

• ming24's FastSpeech2 implementation
• Official DiffSpeech implementation

Citation

@misc{li2023diverse,
  title={Diverse and Expressive Speech Prosody Prediction with Denoising Diffusion Probabilistic Model}, 
  author={Xiang Li and Songxiang Liu and Max W. Y. Lam and Zhiyong Wu and Chao Weng and Helen Meng},
  year={2023},
  eprint={2305.16749},
  archivePrefix={arXiv},
  primaryClass={cs.SD}
}