"A Telephone-Based Speech Recognition System and VoiceXML Application Design Tool"

Supervised by P.Green

Abstract

VoiceXML is a new programming language that makes Internet information and content accessible via voice and phone. By combining state-of-the-art speech technology with the flexibility and portability of XML, VoiceXML has the potential to change the way in which we access the world-wide web.

In this dissertation, a Nuance-based speech recognition system is specially adapted so that it can function over a telephone line. This is then is setup in collaboration with a VoiceXML document server in order to produce a complete system for hosting over-the-phone speech applications. The system is evaluated using a series of isolated word and continuous speech tests.

In the second part of the project, a tool is implemented that allows rapid creation of VoiceXML applications. The tool, Visual VXML, allows users who are unskilled in programming disciplines to construct complex speech applications by virtue of an easy to use visual interface.