Frequency domain TRINICON-based blind source separation method with multi-source activity detection for sparsely mixed signals

TL;DR

This paper enhances frequency domain TRINICON blind source separation by integrating multi-source activity detection to improve separation of sparsely mixed signals, addressing the challenge of intermittent source activity.

Contribution

It introduces a multi-source activity detection method within frequency domain TRINICON to improve separation performance for sparsely mixed signals.

Findings

Improved separation performance demonstrated on sparsely mixed signals.

Regularized filter updating based on activity detection enhances BSS accuracy.

Objective metrics confirm the effectiveness of the proposed method.

Abstract

The TRINICON ('Triple-N ICA for convolutive mixtures') framework is an effective blind signal separation (BSS) method for separating sound sources from convolutive mixtures. It makes full use of the non-whiteness, non-stationarity and non-Gaussianity properties of the source signals and can be implemented either in time domain or in frequency domain, avoiding the notorious internal permutation problem. It usually has best performance when the sources are continuously mixed. In this paper, the offline dual-channel frequency domain TRINICON implementation for sparsely mixed signals is investigated, and a multi-source activity detection is proposed to locate the active period of each source, based on which the filter updating strategy is regularized to improve the separation performance. The objective metric provided by the BSSEVAL toolkit is utilized to evaluate the performance of the proposed scheme.