Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech
This work addresses speech quality assessment for applications like telecommunications or audio processing, but it is incremental as it adapts existing diffusion models to a new task.
The paper tackled the problem of assessing speech quality without requiring annotations by using an unconditional diffusion model trained only on clean speech, and it showed promising results with strong correlation to human scores in listening experiments.
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.