New method improves dysarthric speech assessment using synthesis data

Researchers have developed a method to improve the assessment of dysarthric speech severity by leveraging data from speech synthesis evaluations. This approach uses Mean Opinion Score (MOS) labels from the QualiSpeech corpus to augment limited clinical annotations. Experiments demonstrated that fine-tuning or jointly training models with this synthesis data consistently enhances performance in predicting both intelligibility and naturalness of dysarthric speech, suggesting shared perceptual characteristics between synthesis artifacts and dysarthric speech. AI

IMPACT This research could lead to more scalable and accessible tools for monitoring and analyzing speech disorders like dysarthria.

RANK_REASON Academic paper detailing a new methodology for speech analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• QualiSpeech

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →