Investigation on articulatory and acoustic characteristics of dysarthria

Direct measurement of articulation contributes to figure out the mechanism of dysarthric speech production accurately, comparing to analyzing acoustic signal alone. This study makes a statistical comparison between dysarthric and normal speech, and investigates and analyzes dysarthric characteristics based on articulatory and acoustic data of continuous English utterance. The distribution of the articulatory point is calculated for each vowel, where the vowels show a relatively smaller region for dysarthria than for normal speech. This implies that the speakers with dysarthria may have some difficulties to fully use the articulatory space as the speakers without dysarthria. The rhythm and sound source are analyzed using autocorrelation function, power spectrum and linear prediction coding. The results reveal that the speakers without dysarthria can keep steady rhythm and energy in uttering consonant-vowel repetition sequences but it is hard for the speakers with dysarthria to maintain the stability. Meanwhile, the dysarthric sound source shows unstable period of the fundamental frequency, which reflects that the speakers with dysarthria could not control the glottis movement well.