Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866916793022414848 |
|---|---|
| author | de Oliveira, Danilo Richter, Julius Lemercier, Jean-Marie Welker, Simon Gerkmann, Timo |
| author_facet | de Oliveira, Danilo Richter, Julius Lemercier, Jean-Marie Welker, Simon Gerkmann, Timo |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_17834 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech de Oliveira, Danilo Richter, Julius Lemercier, Jean-Marie Welker, Simon Gerkmann, Timo Audio and Speech Processing Machine Learning Sound 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. |
| title | Non-intrusive Speech Quality Assessment with Diffusion Models Trained on Clean Speech |
| topic | Audio and Speech Processing Machine Learning Sound |
| url | https://arxiv.org/abs/2410.17834 |