Semi-Blind Source Separation for Nonlinear Acoustic Echo Cancellation

TL;DR

This paper introduces a semi-blind source separation approach for nonlinear acoustic echo cancellation that is more robust to model mismatch, improving echo suppression and speech quality.

Contribution

It proposes a semi-blind source separation algorithm based on basis-generic expansion and natural gradient, reducing sensitivity to nonlinear model mismatch in acoustic echo cancellation.

Findings

Outperforms conventional methods in echo return loss enhancement (ERLE)

Improves speech quality as measured by PESQ and STOI

Less sensitive to nonlinear model mismatch

Abstract

The mismatch between the numerical and actual nonlinear models is a challenge to nonlinear acoustic echo cancellation (NAEC) when the nonlinear adaptive filter is utilized. To alleviate this problem, we combine a basis-generic expansion of the memoryless nonlinearity into semi-blind source separation (SBSS). By regarding all the basis functions of the far-end input signal as the known equivalent reference signals, an SBSS updating algorithm is derived following the constrained scaled natural gradient strategy. Unlike the commonly utilized adaptive algorithm, the proposed SBSS is based on the independence between the near-end signal and the reference signals, and is less sensitive to the mismatch of nonlinearity between the numerical and actual models. Experimental results show that the proposed method outperforms conventional methods in terms of echo return loss enhancement (ERLE) and near-end speech quality evaluated by perceptual evaluation of speech quality (PESQ) and short-time objective intelligibility (STOI).