Extract and Diffuse: Latent Integration for Improved Diffusion-based Speech and Vocal Enhancement

TL;DR

This paper introduces Ex-Diff, a novel diffusion model that combines generative and discriminative approaches to enhance speech and vocals, achieving notable improvements in audio quality metrics.

Contribution

The paper presents Ex-Diff, integrating discriminative latent representations into diffusion models for superior speech and vocal enhancement.

Findings

3.7% relative improvement in SI-SDR

10.0% relative improvement in SI-SIR

Demonstrates the complementary strengths of generative and discriminative models

Abstract

Diffusion-based generative models have recently achieved remarkable results in speech and vocal enhancement due to their ability to model complex speech data distributions. While these models generalize well to unseen acoustic environments, they may not achieve the same level of fidelity as the discriminative models specifically trained to enhance particular acoustic conditions. In this paper, we propose Ex-Diff, a novel score-based diffusion model that integrates the latent representations produced by a discriminative model to improve speech and vocal enhancement, which combines the strengths of both generative and discriminative models. Experimental results on the widely used MUSDB dataset show relative improvements of 3.7% in SI-SDR and 10.0% in SI-SIR compared to the baseline diffusion model for speech and vocal enhancement tasks, respectively. Additionally, case studies are provided to further illustrate and analyze the complementary nature of generative and discriminative models in this context.