Improved singing voice separation with chromagram-based pitch-aware remixing

TL;DR

This paper introduces a novel data augmentation method called chromagram-based pitch-aware remixing, which improves singing voice separation performance by better utilizing training data through pitch-aligned mixing.

Contribution

The paper presents a new data augmentation technique that leverages chromagram-based pitch alignment to enhance singing voice separation models.

Findings

Pitch-aware remixing improves SDR in singing voice separation.

Controlled experiments show significant performance gains.

Method benefits both supervised and semi-supervised models.

Abstract

Singing voice separation aims to separate music into vocals and accompaniment components. One of the major constraints for the task is the limited amount of training data with separated vocals. Data augmentation techniques such as random source mixing have been shown to make better use of existing data and mildly improve model performance. We propose a novel data augmentation technique, chromagram-based pitch-aware remixing, where music segments with high pitch alignment are mixed. By performing controlled experiments in both supervised and semi-supervised settings, we demonstrate that training models with pitch-aware remixing significantly improves the test signal-to-distortion ratio (SDR)