High-Quality Visually-Guided Sound Separation from Diverse Categories

TL;DR

DAVIS introduces a diffusion-based generative framework for high-quality audio-visual sound separation, outperforming existing methods by synthesizing sounds directly conditioned on visual and audio cues.

Contribution

The paper presents DAVIS, a novel diffusion model approach for audio-visual separation that surpasses traditional mask-based methods in quality and diversity.

Findings

Outperforms state-of-the-art methods on AVE and MUSIC datasets.

Generates higher-quality separated sounds with diverse categories.

Leverages a diffusion model conditioned on visual and audio data.

Abstract

We propose DAVIS, a Diffusion-based Audio-VIsual Separation framework that solves the audio-visual sound source separation task through generative learning. Existing methods typically frame sound separation as a mask-based regression problem, achieving significant progress. However, they face limitations in capturing the complex data distribution required for high-quality separation of sounds from diverse categories. In contrast, DAVIS leverages a generative diffusion model and a Separation U-Net to synthesize separated sounds directly from Gaussian noise, conditioned on both the audio mixture and the visual information. With its generative objective, DAVIS is better suited to achieving the goal of high-quality sound separation across diverse sound categories. We compare DAVIS to existing state-of-the-art discriminative audio-visual separation methods on the AVE and MUSIC datasets, and results show that DAVIS outperforms other methods in separation quality, demonstrating the advantages of our framework for tackling the audio-visual source separation task.