ISCA Archive Interspeech 2025
ISCA Archive Interspeech 2025

Synthetic Dysarthric Speech: A Supplement, Not a Substitute for Authentic Data in Dysarthric Speech Recognition

Jingting Li, Keyi Feng, Xinran Zhao, Yan Wang, Su-Jing Wang

Dysarthric speech recognition (DSR) is an emerging field that can enhance social interactions and mental health for individuals with dysarthria. However, the lack of sufficient Chinese dysarthric speech data and challenges like ambiguity and individual differences hinder performance improvements. Text-to-speech (TTS) technology is well-established in normal speech recognition and can also supplement dysarthric speech data. This study explores the impact of TTS-based Chinese dysarthric speech generation on DSR performance. Speaker-dependent experiments show that synthetic dysarthric speech alone does not effectively improve DSR performance. Through statistical analysis of acoustic features, we reveal the disparities between synthetic and authentic speech in dysarthria and highlight the limitations of synthetic data for DSR. These findings provide insights for future improvements in speech generation methods.