Speaking technical documents: Using prosody to convey textual and mathematical material

Abstract

Though Braille is the most common means whereby blind people can access information of any kind, it is rapidly being superseded by spoken versions of the same material. Owing to the bulky nature of Braille, the ability to transport a small portable computer, rather than multiple volumes of a book has far greater appeal. However, to date the monotonous nature of synthetic speech has meant that both highly technical information, and the more visually oriented presentational styles (such as mathematics) have been largely inaccessible to blind people. While the ability to approximate human prosody is apparent in some synthesisers, these features are not utilised by the developers of screen-access software. Consequently, the ability to present anything other than purely textual material is distinctly lacking in this type of software. This lack ensures that blind students and professionals working in the scientific or technical arena are to a great extent prevented from reading large amounts of relevant material. This paper describes a model of verbalising mathematics using spoken audio. The language of written mathematics can be translated in to an English representation based on the grammatical structures inherent in the language. The model discussed here encapsulates the structure of an equation in the most intuitive form of communication available; natural speech, while the content is enhanced by the use of alterations in the prosody (inflection) of the voice. It concludes with a discussion of some current areas of investigation. These include the application of certain acoustic affects to the speech signal to convey auditorily, those visual cues so readily apparent from the spatially oriented layout of mathematical content.