Investigating Training Objectives for Generative Speech Enhancement

TL;DR

This paper compares different training objectives for generative speech enhancement, focusing on score-based models and Schr"odinger bridge, and introduces a new perceptual loss to improve speech quality.

Contribution

It provides a detailed comparison of training frameworks and proposes a novel perceptual loss function for Schr"odinger bridge models.

Findings

Schr"odinger bridge models outperform score-based models in speech quality.

The proposed perceptual loss enhances perceptual quality of enhanced speech.

Experimental results are validated with publicly available code and models.

Abstract

Generative speech enhancement has recently shown promising advancements in improving speech quality in noisy environments. Multiple diffusion-based frameworks exist, each employing distinct training objectives and learning techniques. This paper aims to explain the differences between these frameworks by focusing our investigation on score-based generative models and the Schr\"odinger bridge. We conduct a series of comprehensive experiments to compare their performance and highlight differing training behaviors. Furthermore, we propose a novel perceptual loss function tailored for the Schr\"odinger bridge framework, demonstrating enhanced performance and improved perceptual quality of the enhanced speech signals. All experimental code and pre-trained models are publicly available to facilitate further research and development in this domain.