Diffusion-based speech enhancement with a weighted generative-supervised learning loss

TL;DR

This paper introduces a diffusion-based speech enhancement method that combines a generative model with a weighted supervised loss to improve the quality of enhanced speech, showing promising experimental results.

Contribution

The paper proposes augmenting diffusion-based speech enhancement with a weighted MSE loss to better incorporate ground-truth speech information during training.

Findings

Improved speech quality over baseline diffusion models

Effective integration of supervised loss enhances enhancement performance

Experimental results validate the proposed method's effectiveness

Abstract

Diffusion-based generative models have recently gained attention in speech enhancement (SE), providing an alternative to conventional supervised methods. These models transform clean speech training samples into Gaussian noise centered at noisy speech, and subsequently learn a parameterized model to reverse this process, conditionally on noisy speech. Unlike supervised methods, generative-based SE approaches usually rely solely on an unsupervised loss, which may result in less efficient incorporation of conditioned noisy speech. To address this issue, we propose augmenting the original diffusion training objective with a mean squared error (MSE) loss, measuring the discrepancy between estimated enhanced speech and ground-truth clean speech at each reverse process iteration. Experimental results demonstrate the effectiveness of our proposed methodology.