ICML 2020poster132 citations

Non-Autoregressive Neural Text-to-Speech

Kainan Peng, Wei Ping, Zhao Song, Kexin Zhao

Abstract

In this work, we propose ParaNet, a non-autoregressive seq2seq model that converts text to spectrogram. It is fully convolutional and brings 46.7 times speed-up over the lightweight Deep Voice 3 at synthesis, while obtaining reasonably good speech quality. ParaNet also produces stable alignment between text and speech on the challenging test sentences by iteratively improving the attention in a layer-by-layer manner. Furthermore, we build the parallel text-to-speech system by applying various parallel neural vocoders, which can synthesize speech from text through a single feed-forward pass. We also explore a novel VAE-based approach to train the inverse autoregressive flow (IAF) based parallel vocoder from scratch, which avoids the need for distillation from a separately trained WaveNet as previous work.

BibTeX
@InProceedings{pmlr-v119-peng20a,
  title = 	 {Non-Autoregressive Neural Text-to-Speech},
  author =       {Peng, Kainan and Ping, Wei and Song, Zhao and Zhao, Kexin},
  booktitle = 	 {Proceedings of the 37th International Conference on Machine Learning},
  pages = 	 {7586--7598},
  year = 	 {2020},
  editor = 	 {III, Hal Daumé and Singh, Aarti},
  volume = 	 {119},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {13--18 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v119/peng20a/peng20a.pdf},
  url = 	 {https://proceedings.mlr.press/v119/peng20a.html},
  abstract = 	 {In this work, we propose ParaNet, a non-autoregressive seq2seq model that converts text to spectrogram. It is fully convolutional and brings 46.7 times speed-up over the lightweight Deep Voice 3 at synthesis, while obtaining reasonably good speech quality. ParaNet also produces stable alignment between text and speech on the challenging test sentences by iteratively improving the attention in a layer-by-layer manner. Furthermore, we build the parallel text-to-speech system by applying various parallel neural vocoders, which can synthesize speech from text through a single feed-forward pass. We also explore a novel VAE-based approach to train the inverse autoregressive flow (IAF) based parallel vocoder from scratch, which avoids the need for distillation from a separately trained WaveNet as previous work.}
}
Non-Autoregressive Neural Text-to-Speech · ICML 2020