Visual Acoustic Matching

TL;DR

This paper presents a new task called visual acoustic matching, where audio is transformed to match the acoustics of a target environment using visual cues, and introduces a cross-modal transformer model trained with self-supervision.

Contribution

The paper proposes the first approach to visual acoustic matching using a cross-modal transformer and self-supervised learning from in-the-wild videos.

Findings

Outperforms traditional acoustic matching methods.

Successfully translates speech to various real-world environments.

Uses self-supervised training to learn from unlabeled videos.

Abstract

We introduce the visual acoustic matching task, in which an audio clip is transformed to sound like it was recorded in a target environment. Given an image of the target environment and a waveform for the source audio, the goal is to re-synthesize the audio to match the target room acoustics as suggested by its visible geometry and materials. To address this novel task, we propose a cross-modal transformer model that uses audio-visual attention to inject visual properties into the audio and generate realistic audio output. In addition, we devise a self-supervised training objective that can learn acoustic matching from in-the-wild Web videos, despite their lack of acoustically mismatched audio. We demonstrate that our approach successfully translates human speech to a variety of real-world environments depicted in images, outperforming both traditional acoustic matching and more heavily supervised baselines.