Weighted Recursive Least Square Filter and Neural Network based Residual Echo Suppression for the AEC-Challenge

TL;DR

This paper introduces a real-time AEC algorithm combining GCC-PHAT, weighted RLS filtering, and neural network residual suppression, achieving high subjective scores and ranking second in the AEC-Challenge.

Contribution

It proposes a novel semi-blind source separation approach for weighted RLS filtering and integrates neural network-based residual echo suppression for improved performance.

Findings

Achieved a mean subjective score of 4.00

Ranked 2nd in the AEC-Challenge

Demonstrated effective residual echo suppression

Abstract

This paper presents a real-time Acoustic Echo Cancellation (AEC) algorithm submitted to the AEC-Challenge. The algorithm consists of three modules: Generalized Cross-Correlation with PHAse Transform (GCC-PHAT) based time delay compensation, weighted Recursive Least Square (wRLS) based linear adaptive filtering and neural network based residual echo suppression. The wRLS filter is derived from a novel semi-blind source separation perspective. The neural network model predicts a Phase-Sensitive Mask (PSM) based on the aligned reference and the linear filter output. The algorithm achieved a mean subjective score of 4.00 and ranked 2nd in the AEC-Challenge.