Developing a voice user interface with improved usability for people with dysarthria

Abstract

This paper describes the development of a voice user interface (VUI) for Korean users with dysarthria. The development process, from target application decisions to prototype system evaluation, focuses on improving the usability of the interface by reflecting user needs. The first step of development is to decide target VUI application and its functions. 25 dysarthric participants (5 middle school students and 20 adults) are asked to list the devices they want to use with a VUI interface and what purposes they would use VUI devices for. From this user study, SMS sending, web searching and voice dialing on mobile phones and tablet PCs are decided as the target application and its functions. The second step is to design the system of the target application in order to improve usability. 120 people with dysarthria are asked to state the main problems of currently available VUI devices, and it is found that speech recognition failure (88%) is the main problem. This result indicates high speech recognition rate will improve usability. Therefore, to improve the recognition rate, an isolated word recognition based system with a customizable command list and a built-in word prediction function is designed for the target VUI devices. The final step is to develop and evaluate a prototype system. In this study, a prototype is developed for Apple iOS and Android platform devices, and then the system design is modified based on the evaluation results of 5 dysarthric evaluators. © 2012 Springer-Verlag.