Browsing Web Through Audio

VoiceXML (2004) is fundamentally designed to make the Internet content accessible via audio. In other words, a VoiceXML document renders information through sequential audios. Supporting VoiceXML documents, voice browsers allow users to browse the Internet independent of visual attention, and thus make information accessible to visually impaired people. In order to make information universally accessible to all people, Web designers need to present two versions of Web pages, one in the HTML format and the other in the VoiceXML format. Manually maintaining a correspondence between HTML and VoiceXML documents is time-consuming and error-prone. It is, therefore, desirable to automatically transform the representation of Web content from HTML to VoiceXML. With HTML/VoiceXML documents organizing information in a hierarchy, it is feasible and natural to represent their structures using graphs. This abstract introduces a graph-grammar-based approach. Graph grammars, which originated in the late 60s and have stepped into a period with theoretically sound and well-established foundation (Rozenberg, 1997), provide a natural approach to describing a class of graphs sharing common structural properties. In particular, hierarchical structures, e.g. trees, can be intuitively visualized as graphs and clearly specified through graph grammars. Then, the conversion from HTML to VoiceXML can be achieved through graph transformation repetitively from bottom to top