Audio-Visual Speech Separation and Dereverberation with a Two-Stage Multimodal Network

TL;DR

This paper introduces a two-stage multimodal network that combines audio and visual signals to jointly separate and dereverberate speech, significantly improving intelligibility and quality in noisy, reverberant environments.

Contribution

It proposes a novel two-stage multimodal network with joint training and a multi-objective loss for simultaneous speech separation and dereverberation, without needing to know the number of speakers.

Findings

21.10% improvement in ESTOI

0.79 improvement in PESQ

Outperforms several baseline methods

Abstract

Background noise, interfering speech and room reverberation frequently distort target speech in real listening environments. In this study, we address joint speech separation and dereverberation, which aims to separate target speech from background noise, interfering speech and room reverberation. In order to tackle this fundamentally difficult problem, we propose a novel multimodal network that exploits both audio and visual signals. The proposed network architecture adopts a two-stage strategy, where a separation module is employed to attenuate background noise and interfering speech in the first stage and a dereverberation module to suppress room reverberation in the second stage. The two modules are first trained separately, and then integrated for joint training, which is based on a new multi-objective loss function. Our experimental results show that the proposed multimodal network yields consistently better objective intelligibility and perceptual quality than several one-stage and two-stage baselines. We find that our network achieves a 21.10% improvement in ESTOI and a 0.79 improvement in PESQ over the unprocessed mixtures. Moreover, our network architecture does not require the knowledge of the number of speakers.