Front-End for the Oral Control of Applications in Windows Environments

This paper presents the development of an oral interface to control any Windows application by means of speech, providing a user-friendly interface. This front-end is fully configurable using plain text files, being able to manage any program with a graphic environment that works under Windows Operating System, using functions from the Windows API. Hidden Markov Models provide the speech recognition ability, using recursive training of triphone models built with a SpeechDat database. The text to speech system uses an MBROLA based algorithm and it is integrated in a dynamic library. The application is focused on customers with some vision or movement handicap and it is designed to be used in Basque language.