Model-based STFT phase recovery for audio source separation

TL;DR

This paper introduces a novel iterative method for recovering phase in audio source separation by leveraging the temporal continuity of sinusoidal components, outperforming traditional Wiener filter approaches when magnitude estimates are accurate.

Contribution

It proposes a new phase recovery technique based on unwrapping and iterative minimization, improving upon existing methods in audio source separation.

Findings

Outperforms state-of-the-art Wiener filter in experiments

Utilizes sinusoidal phase unwrapping for better phase estimates

Enhances time-domain signal synthesis accuracy

Abstract

For audio source separation applications, it is common to estimate the magnitude of the short-time Fourier transform (STFT) of each source. In order to further synthesizing time-domain signals, it is necessary to recover the phase of the corresponding complex-valued STFT. Most authors in this field choose a Wiener-like filtering approach which boils down to using the phase of the original mixture. In this paper, a different standpoint is adopted. Many music events are partially composed of slowly varying sinusoids and the STFT phase increment over time of those frequency components takes a specific form. This allows phase recovery by an unwrapping technique once a short-term frequency estimate has been obtained. Herein, a novel iterative source separation procedure is proposed which builds upon these results. It consists in minimizing the mixing error by means of the auxiliary function method. This procedure is initialized by exploiting the unwrapping technique in order to generate estimates that benefit from a temporal continuity property. Experiments conducted on realistic music pieces show that, given accurate magnitude estimates, this procedure outperforms the state-of-the-art consistent Wiener filter.