Impairments are Clustered in Latents of Deep Neural Network-based Speech Quality Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866916714819616768 |
|---|---|
| author | Cumlin, Fredrik Liang, Xinyu Ungureanu, Victor Reddy, Chandan K. A. Schüldt, Christian Chatterjee, Saikat |
| author_facet | Cumlin, Fredrik Liang, Xinyu Ungureanu, Victor Reddy, Chandan K. A. Schüldt, Christian Chatterjee, Saikat |
| contents | In this article, we provide an experimental observation: Deep neural network (DNN) based speech quality assessment (SQA) models have inherent latent representations where many types of impairments are clustered. While DNN-based SQA models are not trained for impairment classification, our experiments show good impairment classification results in an appropriate SQA latent representation. We investigate the clustering of impairments using various kinds of audio degradations that include different types of noises, waveform clipping, gain transition, pitch shift, compression, reverberation, etc. To visualize the clusters we perform classification of impairments in the SQA-latent representation domain using a standard k-nearest neighbor (kNN) classifier. We also develop a new DNN-based SQA model, named DNSMOS+, to examine whether an improvement in SQA leads to an improvement in impairment classification. The classification accuracy is 94% for LibriAugmented dataset with 16 types of impairments and 54% for ESC-50 dataset with 50 types of real noises. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_21528 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Impairments are Clustered in Latents of Deep Neural Network-based Speech Quality Models Cumlin, Fredrik Liang, Xinyu Ungureanu, Victor Reddy, Chandan K. A. Schüldt, Christian Chatterjee, Saikat Audio and Speech Processing In this article, we provide an experimental observation: Deep neural network (DNN) based speech quality assessment (SQA) models have inherent latent representations where many types of impairments are clustered. While DNN-based SQA models are not trained for impairment classification, our experiments show good impairment classification results in an appropriate SQA latent representation. We investigate the clustering of impairments using various kinds of audio degradations that include different types of noises, waveform clipping, gain transition, pitch shift, compression, reverberation, etc. To visualize the clusters we perform classification of impairments in the SQA-latent representation domain using a standard k-nearest neighbor (kNN) classifier. We also develop a new DNN-based SQA model, named DNSMOS+, to examine whether an improvement in SQA leads to an improvement in impairment classification. The classification accuracy is 94% for LibriAugmented dataset with 16 types of impairments and 54% for ESC-50 dataset with 50 types of real noises. |
| title | Impairments are Clustered in Latents of Deep Neural Network-based Speech Quality Models |
| topic | Audio and Speech Processing |
| url | https://arxiv.org/abs/2504.21528 |