SqueezeWave: Extremely Lightweight Vocoders for On-device Speech Synthesis

ArXiv: arXiv

Github: Click me

Authors

By Tianren Gao *, Bohan Zhai *, Flora Xue, Daniel Rothchild, Bichen Wu, Joseph Gonzalez, and Kurt Keutzer (UC Berkeley)

* Equal contribution.

Abstract

Automatic speech synthesis is a challenging task that is becoming increasingly important as edge devices begin to interact with users through speech. Typical text-to-speech pipelines include a vocoder, which translates intermediate audio representations into an audio waveform. Most existing vocoders are difficult to parallelize since each generated sample is conditioned on previous samples. WaveGlow is a flow-based feed-forward alternative to these auto-regressive models (Prenger et al., 2019). However, while WaveGlow can be easily parallelized, the model is too expensive for real-time speech synthesis on the edge. This paper presents SqueezeWave, a family of lightweight vocoders based on WaveGlow that can generate audio of similar quality to WaveGlow with 61x - 214x fewer MACs. Code, trained models, and generated audio are publicly available at: Link

LJ0015

LJ0051

LJ0063

LJ0072

LJ0079

LJ0094

LJ0096

LJ0102

LJ0153

LJ00173