Enhancing Dysarthric Speech Recognition for Unseen Speakers via Prototype-Based Adaptation

TL;DR

This paper proposes a prototype-based method using HuBERT features and contrastive learning to improve dysarthric speech recognition for unseen speakers without fine-tuning, enhancing personalization and performance.

Contribution

Introduces a novel prototype-based adaptation approach that leverages HuBERT features and contrastive learning to improve recognition accuracy for new dysarthric speakers without additional model fine-tuning.

Findings

Significant performance gains on unseen speakers

Effective personalization without fine-tuning

Prototypes capture speaker-specific speech characteristics

Abstract

Dysarthric speech recognition (DSR) presents a formidable challenge due to inherent inter-speaker variability, leading to severe performance degradation when applying DSR models to new dysarthric speakers. Traditional speaker adaptation methodologies typically involve fine-tuning models for each speaker, but this strategy is cost-prohibitive and inconvenient for disabled users, requiring substantial data collection. To address this issue, we introduce a prototype-based approach that markedly improves DSR performance for unseen dysarthric speakers without additional fine-tuning. Our method employs a feature extractor trained with HuBERT to produce per-word prototypes that encapsulate the characteristics of previously unseen speakers. These prototypes serve as the basis for classification. Additionally, we incorporate supervised contrastive learning to refine feature extraction. By enhancing representation quality, we further improve DSR performance, enabling effective personalized DSR. We release our code at https://github.com/NKU-HLT/PB-DSR.