Dynamic Aural Browsing of MathML Documents via VoiceXML

Understanding and internalizing complex mathematical expressions is a difficult task for visually impaired individuals. For visually impaired individuals, understanding mathematical expressions entails being able to “browse” the expression, i.e., navigating the expression and hearing the whole expression or its sub-parts repeatedly. Designing aural browsers that allow such navigation is a difficult task. In this paper we present a VoiceXML-based approach for aurally navigating mathematical expressions coded in MathML. In our approach a MathML document is translated into a VoiceXML document via the XSLT transformation language. The VoiceXML document is further augmented so that users can attach speech labels (termed voice anchors by us) at various points in the document while browsing. These speech labels or voice anchors can then be used by the listener to dynamically navigate the expression during the browsing session. A prototype system based on these ideas has been implemented and tested.