End-to-End Complex-Valued Multidilated Convolutional Neural Network for Joint Acoustic Echo Cancellation and Noise Suppression

TL;DR

This paper introduces an end-to-end complex-valued neural network that jointly performs acoustic echo cancellation and noise suppression, eliminating the need for separate adaptive filtering modules and reducing delays in full-duplex communication systems.

Contribution

It proposes a novel pseudocomplex extension of D3Net for joint echo and noise suppression, with a dual-mask technique and a small, efficient architecture that leverages multi-resolution features.

Findings

Achieved promising results on synthetic and real test sets.

Outperformed traditional two-stage frameworks in reducing delays.

Demonstrated effectiveness across multiple metrics.

Abstract

Echo and noise suppression is an integral part of a full-duplex communication system. Many recent acoustic echo cancellation (AEC) systems rely on a separate adaptive filtering module for linear echo suppression and a neural module for residual echo suppression. However, not only do adaptive filtering modules require convergence and remain susceptible to changes in acoustic environments, but this two-stage framework also often introduces unnecessary delays to the AEC system when neural modules are already capable of both linear and nonlinear echo suppression. In this paper, we exploit the offset-compensating ability of complex time-frequency masks and propose an end-to-end complex-valued neural network architecture. The building block of the proposed model is a pseudocomplex extension based on the densely-connected multidilated DenseNet (D3Net) building block, resulting in a very small network of only 354K parameters. The architecture utilized the multi-resolution nature of the D3Net building blocks to eliminate the need for pooling, allowing the network to extract features using large receptive fields without any loss of output resolution. We also propose a dual-mask technique for joint echo and noise suppression with simultaneous speech enhancement. Evaluation on both synthetic and real test sets demonstrated promising results across multiple energy-based metrics and perceptual proxies.