SqueezeWave: Extremely Lightweight Vocoders for On-device Speech Synthesis

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 https://github.com/tianrengao/SqueezeWave.

References in corpus (5)

• MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
• SqueezeDet: Unified, Small, Low Power Fully Convolutional Neural Networks for Real-Time Object Detection for Autonomous Driving
• Efficient Neural Audio Synthesis
• Shift: A Zero FLOP, Zero Parameter Alternative to Spatial Convolutions
• SqueezeSegV2: Improved Model Structure and Unsupervised Domain Adaptation for Road-Object Segmentation from a LiDAR Point Cloud

Cited by in corpus (8)

• HiFi-GAN: Generative Adversarial Networks for Efficient and High Fidelity Speech Synthesis
• A Survey on Neural Speech Synthesis
• A Flow-Based Neural Network for Time Domain Speech Enhancement
• Speech Synthesis and Control Using Differentiable DSP
• Parallel Synthesis for Autoregressive Speech Generation
• ConVoice: Real-Time Zero-Shot Voice Style Transfer with Convolutional Network
• FBWave: Efficient and Scalable Neural Vocoders for Streaming Text-To-Speech on the Edge
• Text-to-speech for the hearing impaired