MoMuSE: Momentum Multi-modal Target Speaker Extraction for Real-time Scenarios with Impaired Visual Cues

TL;DR

MoMuSE is a real-time multi-modal speaker extraction method that maintains speaker identity momentum to improve target speech extraction when visual cues are impaired.

Contribution

It introduces a novel momentum-based memory mechanism for continuous speaker tracking in audio-visual extraction under visual impairment conditions.

Findings

Significant improvement in scenarios with impaired visual cues

Effective real-time inference with momentum memory

Enhanced robustness over traditional AV-TSE methods

Abstract

Audio-visual Target Speaker Extraction (AV-TSE) aims to isolate the speech of a specific target speaker from an audio mixture using time-synchronized visual cues. In real-world scenarios, visual cues are not always available due to various impairments, which undermines the stability of AV-TSE. Despite this challenge, humans can maintain attentional momentum over time, even when the target speaker is not visible. In this paper, we introduce the Momentum Multi-modal target Speaker Extraction (MoMuSE), which retains a speaker identity momentum in memory, enabling the model to continuously track the target speaker. Designed for real-time inference, MoMuSE extracts the current speech window with guidance from both visual cues and dynamically updated speaker momentum. Experimental results demonstrate that MoMuSE exhibits significant improvement, particularly in scenarios with severe impairment of visual cues.