Listener-Controlled Dynamic Navigation of VoiceXML Documents

The next frontier for research on the Web is to make it universally accessible, primarily via audio. Work in this direction includes the development of voice browsers and the design of VoiceXML, an XML-based standard for marking-up documents to be played on voice browsers. VoiceXML is interactive and allows voice input to be received and processed by a voice browser (i.e., VoiceXML documents can be aurally navigated). Unfortunately, this interaction is completely controlled by the author of the VoiceXML document. Coupled with the sequential nature of audio output, this results in a navigation process that is tedious and not under the listener’s control. It also inhibits the development of more advanced applications that require aural navigation. In this paper we develop techniques that put aural navigation under the listener’s control. We propose voice anchors that allow listeners to aurally tag specific points in the audio document for access via spoken commands later. We discuss a spectrum of techniques for realizing anchors and their implementation. We also introduce voice-commanded scripting languages for orally programming aural navigation strategies.