Egocentric Audio-Visual Noise Suppression

TL;DR

This paper introduces a novel egocentric audio-visual noise suppression method that leverages visual cues from non-lip facial visuals and actions, employing multi-task learning to improve noise reduction in videos where the speaker is off-screen.

Contribution

It proposes a new framework that uses visual features for noise suppression in egocentric videos, including multi-task learning for better noise and event discrimination.

Findings

Visual features improve noise suppression performance.

Multi-task learning enhances noise reduction and acoustic event detection.

The model outperforms audio-only baselines across various conditions.

Abstract

This paper studies audio-visual noise suppression for egocentric videos -- where the speaker is not captured in the video. Instead, potential noise sources are visible on screen with the camera emulating the off-screen speaker's view of the outside world. This setting is different from prior work in audio-visual speech enhancement that relies on lip and facial visuals. In this paper, we first demonstrate that egocentric visual information is helpful for noise suppression. We compare object recognition and action classification-based visual feature extractors and investigate methods to align audio and visual representations. Then, we examine different fusion strategies for the aligned features, and locations within the noise suppression model to incorporate visual information. Experiments demonstrate that visual features are most helpful when used to generate additive correction masks. Finally, in order to ensure that the visual features are discriminative with respect to different noise types, we introduce a multi-task learning framework that jointly optimizes audio-visual noise suppression and video-based acoustic event detection. This proposed multi-task framework outperforms the audio-only baseline on all metrics, including a 0.16 PESQ improvement. Extensive ablations reveal the improved performance of the proposed model with multiple active distractors, overall noise types, and across different SNRs.