Self-Supervised VQ-VAE for One-Shot Music Style Transfer

TL;DR

This paper introduces a novel self-supervised VQ-VAE-based method for one-shot music style transfer, specifically targeting timbre transfer, and demonstrates its superior performance over existing baselines.

Contribution

The work extends VQ-VAE with a self-supervised learning strategy to achieve disentangled representations for one-shot timbre transfer in music.

Findings

Outperforms baseline methods on objective metrics

Successfully disentangles timbre and pitch representations

Effective in one-shot music style transfer tasks

Abstract

Neural style transfer, allowing to apply the artistic style of one image to another, has become one of the most widely showcased computer vision applications shortly after its introduction. In contrast, related tasks in the music audio domain remained, until recently, largely untackled. While several style conversion methods tailored to musical signals have been proposed, most lack the 'one-shot' capability of classical image style transfer algorithms. On the other hand, the results of existing one-shot audio style transfer methods on musical inputs are not as compelling. In this work, we are specifically interested in the problem of one-shot timbre transfer. We present a novel method for this task, based on an extension of the vector-quantized variational autoencoder (VQ-VAE), along with a simple self-supervised learning strategy designed to obtain disentangled representations of timbre and pitch. We evaluate the method using a set of objective metrics and show that it is able to outperform selected baselines.