WaveFlow: A Compact Flow-based Model for Raw Audio

Wei Ping, Kainan Peng, Kexin Zhao, Zhao Song
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:7706-7716, 2020.

Abstract

In this work, we propose WaveFlow, a small-footprint generative flow for raw audio, which is directly trained with maximum likelihood. It handles the long-range structure of 1-D waveform with a dilated 2-D convolutional architecture, while modeling the local variations using expressive autoregressive functions. WaveFlow provides a unified view of likelihood-based models for 1-D data, including WaveNet and WaveGlow as special cases. It generates high-fidelity speech as WaveNet, while synthesizing several orders of magnitude faster as it only requires a few sequential steps to generate very long waveforms with hundreds of thousands of time-steps. Furthermore, it can significantly reduce the likelihood gap that has existed between autoregressive models and flow-based models for efficient synthesis. Finally, our small-footprint WaveFlow has only 5.91M parameters, which is 15{\texttimes} smaller than WaveGlow. It can generate 22.05 kHz high-fidelity audio 42.6{\texttimes} faster than real-time (at a rate of 939.3 kHz) on a V100 GPU without engineered inference kernels.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-ping20a, title = {{W}ave{F}low: A Compact Flow-based Model for Raw Audio}, author = {Ping, Wei and Peng, Kainan and Zhao, Kexin and Song, Zhao}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {7706--7716}, 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/ping20a/ping20a.pdf}, url = {http://proceedings.mlr.press/v119/ping20a.html}, abstract = {In this work, we propose WaveFlow, a small-footprint generative flow for raw audio, which is directly trained with maximum likelihood. It handles the long-range structure of 1-D waveform with a dilated 2-D convolutional architecture, while modeling the local variations using expressive autoregressive functions. WaveFlow provides a unified view of likelihood-based models for 1-D data, including WaveNet and WaveGlow as special cases. It generates high-fidelity speech as WaveNet, while synthesizing several orders of magnitude faster as it only requires a few sequential steps to generate very long waveforms with hundreds of thousands of time-steps. Furthermore, it can significantly reduce the likelihood gap that has existed between autoregressive models and flow-based models for efficient synthesis. Finally, our small-footprint WaveFlow has only 5.91M parameters, which is 15{\texttimes} smaller than WaveGlow. It can generate 22.05 kHz high-fidelity audio 42.6{\texttimes} faster than real-time (at a rate of 939.3 kHz) on a V100 GPU without engineered inference kernels.} }
Endnote
%0 Conference Paper %T WaveFlow: A Compact Flow-based Model for Raw Audio %A Wei Ping %A Kainan Peng %A Kexin Zhao %A Zhao Song %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-ping20a %I PMLR %P 7706--7716 %U http://proceedings.mlr.press/v119/ping20a.html %V 119 %X In this work, we propose WaveFlow, a small-footprint generative flow for raw audio, which is directly trained with maximum likelihood. It handles the long-range structure of 1-D waveform with a dilated 2-D convolutional architecture, while modeling the local variations using expressive autoregressive functions. WaveFlow provides a unified view of likelihood-based models for 1-D data, including WaveNet and WaveGlow as special cases. It generates high-fidelity speech as WaveNet, while synthesizing several orders of magnitude faster as it only requires a few sequential steps to generate very long waveforms with hundreds of thousands of time-steps. Furthermore, it can significantly reduce the likelihood gap that has existed between autoregressive models and flow-based models for efficient synthesis. Finally, our small-footprint WaveFlow has only 5.91M parameters, which is 15{\texttimes} smaller than WaveGlow. It can generate 22.05 kHz high-fidelity audio 42.6{\texttimes} faster than real-time (at a rate of 939.3 kHz) on a V100 GPU without engineered inference kernels.
APA
Ping, W., Peng, K., Zhao, K. & Song, Z.. (2020). WaveFlow: A Compact Flow-based Model for Raw Audio. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:7706-7716 Available from http://proceedings.mlr.press/v119/ping20a.html.

Related Material