Efficient Neural Audio Synthesis

TL;DR

This paper introduces WaveRNN, a compact and efficient neural network for high-quality audio synthesis that significantly reduces sampling time, employs weight pruning for sparsity, and uses subscaling for parallel sample generation.

Contribution

The paper presents WaveRNN with a dual softmax layer, applies weight pruning for sparse networks, and introduces subscaling for parallel sample generation, advancing efficient neural audio synthesis.

Findings

WaveRNN matches WaveNet quality with 4x faster GPU synthesis.

Sparse WaveRNN outperforms dense networks at the same parameter count.

Subscale WaveRNN enables parallel sample generation without quality loss.

Abstract

Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24kHz 16-bit audio 4x faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency.