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

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 severa...

Full description

Saved in:
Bibliographic Details
Main Authors: Cifka, Ondrej, Ozerov, Alexey, Simsekli, Umut, Richard, Gael
Format: Conference Proceeding
Language:English
Subjects:
Computer vision
Conferences
Decoding
deep learning
Instruments
Measurement
music
self-supervised learning
Signal processing
Signal processing algorithms
Style transfer
timbre
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary: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.
ISSN:2379-190X
DOI:10.1109/ICASSP39728.2021.9414235