Enregistré dans:
Détails bibliographiques
Auteurs principaux: Yang, Jianing, Nakata, Wataru, Saito, Yuki, Saruwatari, Hiroshi
Format: Preprint
Publié: 2026
Sujets:
Accès en ligne:https://arxiv.org/abs/2601.13700
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866908777000730624
author Yang, Jianing
Nakata, Wataru
Saito, Yuki
Saruwatari, Hiroshi
author_facet Yang, Jianing
Nakata, Wataru
Saito, Yuki
Saruwatari, Hiroshi
contents With the advancement of self-supervised learning (SSL), fine-tuning pretrained SSL models for mean opinion score (MOS) prediction has achieved state-of-the-art performance. However, during fine-tuning, these SSL-based MOS prediction models often suffer from catastrophic forgetting of the pretrained knowledge and tend to overfit the training set, resulting in poor generalization performance. In this study, we propose DistilMOS, a novel method that learns to predict not only MOS but also token IDs obtained by clustering the hidden representations of each layer in the pretrained SSL model. These layer-wise token targets serve as self-distillation signals that enables the MOS prediction model to extract rich internal knowledge from SSL models, enhancing both prediction accuracy and generalization capability. Experimental evaluations demonstrate that our method significantly outperforms standard SSL-based MOS prediction models on both in-domain and out-of-domain evaluations, verifying the effectiveness and practicality of the proposed method.
format Preprint
id arxiv_https___arxiv_org_abs_2601_13700
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle DistilMOS: Layer-Wise Self-Distillation For Self-Supervised Learning Model-Based MOS Prediction
Yang, Jianing
Nakata, Wataru
Saito, Yuki
Saruwatari, Hiroshi
Sound
With the advancement of self-supervised learning (SSL), fine-tuning pretrained SSL models for mean opinion score (MOS) prediction has achieved state-of-the-art performance. However, during fine-tuning, these SSL-based MOS prediction models often suffer from catastrophic forgetting of the pretrained knowledge and tend to overfit the training set, resulting in poor generalization performance. In this study, we propose DistilMOS, a novel method that learns to predict not only MOS but also token IDs obtained by clustering the hidden representations of each layer in the pretrained SSL model. These layer-wise token targets serve as self-distillation signals that enables the MOS prediction model to extract rich internal knowledge from SSL models, enhancing both prediction accuracy and generalization capability. Experimental evaluations demonstrate that our method significantly outperforms standard SSL-based MOS prediction models on both in-domain and out-of-domain evaluations, verifying the effectiveness and practicality of the proposed method.
title DistilMOS: Layer-Wise Self-Distillation For Self-Supervised Learning Model-Based MOS Prediction
topic Sound
url https://arxiv.org/abs/2601.13700