STARDUST – Speech Training And Recognition for Dysarthric Users of Assistive Technology

Automatic speech recognition (ASR) can provide a rapid means of controlling EAT. Off-the-shelf ASR systems function poorly for users with severe dysarthria because of the increased variability of their articulations compared to ‘normal’ speech. A two-pronged approach has been applied to this problem: 1. To develop a computerised training package which will assist dysarthric speakers to improve the recognition likelihood and consistency of their vocalisations. 2. To develop speech recognition systems which have greater tolerance to variability of speech utterances. We present results of trials to evaluate the effect of the speech training aid on the speech of dysarthric individuals. Initial results have shown good speech recognition rates for people with even the most severe dysarthria. Speech command driven environmental control systems and voice output communication aids are being developed.