Acoustic echo suppression using a learning-based multi-frame minimum variance distortionless response filter

TL;DR

This paper introduces a learning-based multi-frame MFMVDR filter for acoustic echo suppression that improves speech quality without double-talk detection, using deep learning for parameter estimation and matrix inversion.

Contribution

It extends the MFMVDR filter with deep learning to directly estimate the inverse correlation matrix, enhancing stability and eliminating the need for double-talk detection.

Findings

Effective in double-talk scenarios

Reduces background noise and nonlinear distortion

Improves speech quality in simulations

Abstract

Distortion resulting from acoustic echo suppression (AES) is a common issue in full-duplex communication. To address the distortion problem, a multi-frame minimum variance distortionless response (MFMVDR) filtering technique is proposed. The MFMVDR filter with parameter estimation which was used in speech enhancement problems is extended in this study from a deep learning perspective. To alleviate numerical instability of the MFMVDR filter, we propose to directly estimate the inverse of the correlation matrix. The AES system is advantageous in that no double-talk detection is required. The negative scale-invariant signal-to-distortion ratio is employed as the loss function in training the network at the output of the MFMVDR filter. Simulation results have demonstrated the efficacy of the proposed learning-based AES system in double-talk, background noise, and nonlinear distortion conditions.