Research

Paper

TESTING March 16, 2026

Something from Nothing: Data Augmentation for Robust Severity Level Estimation of Dysarthric Speech

Authors

Jaesung Bae, Xiuwen Zheng, Minje Kim, Chang D. Yoo, Mark Hasegawa-Johnson

Abstract

Dysarthric speech quality assessment (DSQA) is critical for clinical diagnostics and inclusive speech technologies. However, subjective evaluation is costly and difficult to scale, and the scarcity of labeled data limits robust objective modeling. To address this, we propose a three-stage framework that leverages unlabeled dysarthric speech and large-scale typical speech datasets to scale training. A teacher model first generates pseudo-labels for unlabeled samples, followed by weakly supervised pretraining using a label-aware contrastive learning strategy that exposes the model to diverse speakers and acoustic conditions. The pretrained model is then fine-tuned for the downstream DSQA task. Experiments on five unseen datasets spanning multiple etiologies and languages demonstrate the robustness of our approach. Our Whisper-based baseline significantly outperforms SOTA DSQA predictors such as SpICE, and the full framework achieves an average SRCC of 0.761 across unseen test datasets.

Metadata

arXiv ID: 2603.15988
Provider: ARXIV
Primary Category: eess.AS
Published: 2026-03-16
Fetched: 2026-03-18 06:02

Related papers

Raw Data (Debug)
{
  "raw_xml": "<entry>\n    <id>http://arxiv.org/abs/2603.15988v1</id>\n    <title>Something from Nothing: Data Augmentation for Robust Severity Level Estimation of Dysarthric Speech</title>\n    <updated>2026-03-16T23:00:07Z</updated>\n    <link href='https://arxiv.org/abs/2603.15988v1' rel='alternate' type='text/html'/>\n    <link href='https://arxiv.org/pdf/2603.15988v1' rel='related' title='pdf' type='application/pdf'/>\n    <summary>Dysarthric speech quality assessment (DSQA) is critical for clinical diagnostics and inclusive speech technologies. However, subjective evaluation is costly and difficult to scale, and the scarcity of labeled data limits robust objective modeling. To address this, we propose a three-stage framework that leverages unlabeled dysarthric speech and large-scale typical speech datasets to scale training. A teacher model first generates pseudo-labels for unlabeled samples, followed by weakly supervised pretraining using a label-aware contrastive learning strategy that exposes the model to diverse speakers and acoustic conditions. The pretrained model is then fine-tuned for the downstream DSQA task. Experiments on five unseen datasets spanning multiple etiologies and languages demonstrate the robustness of our approach. Our Whisper-based baseline significantly outperforms SOTA DSQA predictors such as SpICE, and the full framework achieves an average SRCC of 0.761 across unseen test datasets.</summary>\n    <category scheme='http://arxiv.org/schemas/atom' term='eess.AS'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.AI'/>\n    <category scheme='http://arxiv.org/schemas/atom' term='cs.LG'/>\n    <published>2026-03-16T23:00:07Z</published>\n    <arxiv:comment>Submitted to Interspeech 2026</arxiv:comment>\n    <arxiv:primary_category term='eess.AS'/>\n    <author>\n      <name>Jaesung Bae</name>\n    </author>\n    <author>\n      <name>Xiuwen Zheng</name>\n    </author>\n    <author>\n      <name>Minje Kim</name>\n    </author>\n    <author>\n      <name>Chang D. Yoo</name>\n    </author>\n    <author>\n      <name>Mark Hasegawa-Johnson</name>\n    </author>\n  </entry>"
}