Dysarthria

About: Dysarthria is a research topic. Over the lifetime, 2402 publications have been published within this topic receiving 56554 citations.

Articulatory movements during vowels produced by speakers with dysarthria and normal controls

Abstract: The central goal of the study was to provide a quantitative description of contributions of the jaw, lower lip, tongue blade and tongue dorsum to vowel productions, and to determine patterns of interarticulatory interactions between movements. Kinematic and acoustic signals were collected using the x‐ray microbeam. Thirty‐four speakers, 12 with dysarthria due to Parkinson disease, 7 with amyotrophic lateral sclerosis, and 15 normal controls, were recorded reading sentences at a comfortable speech rate. Ten CVC words, each containing one of the English vowels /i,I,u,a,ae,o/ carrying primary stress, were selected for analysis. Each fleshpoint trajectory was characterized by marker positions at vowel onset and offset, and the moment when speed was lowest. Measures of distance traveled, time to and from the moment of minimum speed, and peak and average movement speed were employed. Movement characteristics, and associations between movements, were compared for different vowels, contexts, speakers and groups. Results are reported for vowels and vowel groups (e.g., lax versus tense), averaged separately by contexts for speaker groups. The data speak to previous claims that speakers with dysarthria exhibit evidence of discoordination in speech movements relative to normal performance. [Work supported by NIDCD Award R01 DC003723.]

Multi-Modal Acoustic-Articulatory Feature Fusion For Dysarthric Speech Recognition

Abstract: Building automatic speech recognition (ASR) systems for speakers with dysarthria is a very challenging task. Although multi-modal ASR has received increasing attention recently, incorporating real articulatory data with acoustic features has not been widely explored in the dysarthric speech community. This paper investigates the effectiveness of multi-modal acoustic modelling for dysarthric speech recognition using acoustic features along with articulatory information. The proposed multi-stream architectures consist of convolutional, recurrent and fully-connected layers allowing for bespoke per-stream pre-processing, fusion at the optimal level of abstraction and post-processing. We study the optimal fusion level/scheme as well as training dynamics in terms of cross-entropy and WER using the popular TORGO dysarthric speech database. Experimental results show that fusing the acoustic and articulatory features at the empirically found optimal level of abstraction achieves a remarkable performance gain, leading to up to 4.6% absolute (9.6% relative) WER reduction for speakers with dysarthria.