Plug-and-Steer: Decoupling Separation and Selection in Audio-Visual Target Speaker Extraction

TL;DR

This paper introduces Plug-and-Steer, a novel approach for audio-visual target speaker extraction that decouples separation and target selection, improving fidelity and flexibility by leveraging a frozen audio backbone and a minimal visual influence.

Contribution

It proposes the Latent Steering Matrix to re-route features, enabling high-fidelity separation with diverse architectures while isolating target selection to visual cues.

Findings

Effective preservation of acoustic priors across architectures

Achieves perceptual quality comparable to original backbones

Enhances robustness in noisy, real-world scenarios

Abstract

The goal of this paper is to provide a new perspective on audio-visual target speaker extraction (AV-TSE) by decoupling the separation and target selection. Conventional AV-TSE systems typically integrate audio and visual features deeply to re-learn the entire separation process, which can act as a fidelity ceiling due to the noisy nature of in-the-wild audio-visual datasets. To address this, we propose Plug-and-Steer, which assigns high-fidelity separation to a frozen audio-only backbone and limits the role of visual modality strictly to target selection. We introduce the Latent Steering Matrix (LSM), a minimalist linear transformation that re-routes latent features within the backbone to anchor the target speaker to a designated channel. Experiments across four representative architectures show that our method effectively preserves the acoustic priors of diverse backbones, achieving perceptual quality comparable to the original backbones. Audio samples are available at: https://plugandsteer.github.io