A Lightweight Hybrid Dual Channel Speech Enhancement System under Low-SNR Conditions

TL;DR

This paper introduces a lightweight hybrid dual-channel speech enhancement system that combines IVA and a modified GTCRN to improve speech quality under low-SNR conditions with minimal computational resources.

Contribution

It presents a novel hybrid system integrating IVA and a modified GTCRN for efficient speech enhancement in resource-constrained, low-SNR environments.

Findings

Effective speech enhancement with minimal parameters

Low computational complexity achieved

Improved speech quality in low-SNR conditions

Abstract

Although deep learning based multi-channel speech enhancement has achieved significant advancements, its practical deployment is often limited by constrained computational resources, particularly in low signal-to-noise ratio (SNR) conditions. In this paper, we propose a lightweight hybrid dual-channel speech enhancement system that combines independent vector analysis (IVA) with a modified version of the dual-channel grouped temporal convolutional recurrent network (GTCRN). IVA functions as a coarse estimator, providing auxiliary information for both speech and noise, while the modified GTCRN further refines the speech quality. We investigate several modifications to ensure the comprehensive utilization of both original and auxiliary information. Experimental results demonstrate the effectiveness of the proposed system, achieving enhanced speech with minimal parameters and low computational complexity.