Voice input for the disabled

Abstract

This thesis reports on the implementation and assessment of a commercially available voice recogniser in practical situations within the disabled community served by the Regency Park Centre for the Young Disabled. Initially a background is provided which illustrates the relevance and importance of technical aids, such as voice input equipment, for the disabled. A consideration of the development of the technology, the principles of operation and the needs of the population served by Regency Park leads to particular applications which are investigated. These applications are in the areas of data entry by voice and articulation training. Initial entry by testing confirmed the potential of the equipment for data voice. Software was developed which made the equipment more suitable for classroom use. Trials involving disabled children were carried out in the laboratory and in the classroom. It was found that, for the trial group of children, other modalities offered greater satisfaction than voice input. Data gathered during these trials is presented as a guide to the performance of the equipment. The data is also used to illustrate a proposed methodology for the objective comparison of interfaces between the disabled and their technical aids. The second application considered was articulation training using the voice recogniser. In this area the work centred on establishing whether the voice recogniser discriminated between sounds with sufficient accuracy to be used for articulation training. Software was developed which would be suitable for articulation training of young children. Experimental work was carried out using the software with able bodied adults saying a series of sounds simulating common pronunciation errors. The results indicated that the performance of the voice recogniser varied greatly from person to person. The data tended to suggest that the variations were made up of a relatively fixed level of substitution errors and a variable quantity of rejection errors which tend to diminish with practise. It is suggested that this characteristic may allow the equipment to be used successfully for the improvement of consistency in articulation, however further clinical testing would be required to demonstrate this.