Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech

TL;DR

This paper introduces a novel unsupervised speech quality assessment method using diffusion models trained solely on clean speech data, which correlates well with human judgments without requiring annotated datasets.

Contribution

It is the first to utilize diffusion models for density estimation in speech quality assessment, enabling unsupervised evaluation based on clean speech priors.

Findings

Log-likelihoods correlate strongly with intrusive speech quality metrics.

The method achieves the best correlation with human scores in listening tests.

It operates without supervision or annotated data.

Abstract

Diffusion models have found great success in generating high quality, natural samples of speech, but their potential for density estimation for speech has so far remained largely unexplored. In this work, we leverage an unconditional diffusion model trained only on clean speech for the assessment of speech quality. We show that the quality of a speech utterance can be assessed by estimating the likelihood of a corresponding sample in the terminating Gaussian distribution, obtained via a deterministic noising process. The resulting method is purely unsupervised, trained only on clean speech, and therefore does not rely on annotations. Our diffusion-based approach leverages clean speech priors to assess quality based on how the input relates to the learned distribution of clean data. Our proposed log-likelihoods show promising results, correlating well with intrusive speech quality metrics and showing the best correlation with human scores in a listening experiment.