SVSGAN: Singing Voice Separation via Generative Adversarial Network

TL;DR

This paper introduces SVSGAN, a novel GAN-based framework for singing voice separation that improves performance by combining supervised initialization with unsupervised adversarial training on multiple datasets.

Contribution

The paper proposes a new GAN-based approach for singing voice separation that leverages distribution matching and combines supervised and unsupervised training phases.

Findings

Improved separation performance on MIR-1K, iKala, and DSD100 datasets.

Effective use of GAN for audio source separation.

Outperforms some existing deep learning methods.

Abstract

Separating two sources from an audio mixture is an important task with many applications. It is a challenging problem since only one signal channel is available for analysis. In this paper, we propose a novel framework for singing voice separation using the generative adversarial network (GAN) with a time-frequency masking function. The mixture spectra is considered to be a distribution and is mapped to the clean spectra which is also considered a distribtution. The approximation of distributions between mixture spectra and clean spectra is performed during the adversarial training process. In contrast with current deep learning approaches for source separation, the parameters of the proposed framework are first initialized in a supervised setting and then optimized by the training procedure of GAN in an unsupervised setting. Experimental results on three datasets (MIR-1K, iKala and DSD100) show that performance can be improved by the proposed framework consisting of conventional networks.