Towards accessible technical documents: production of speech and Braille output from formatted documents

摘要

The primary objective of this research, was to devise methods for communicating highly technical material to blind people, through the medium of Braille or prosodically enhanced spoken output. This solution necessitated devising strategies to both model the document internally, and to unambiguously produce the material in the two output media.ududThe first phase m the generation of intelligible output was the transformation of the LATEX source into well-formatted and accurate Braille. Following on from this, methodologies were defined to convey the structure and textual content of documents using prosodic alterations to the synthetic voice. We have devised mechamsms whereby mathematical content can be delivered in an intuitive manner, using the sole medium of prosodically enhanced spoken output. This ensures that the listener will not have to learn specific non-speech auditory sound to gam access to this form of presentation.ududWe have also devised a newer, and more flexible means for representing the structure and content of the document m the computer’s memory. This Directed Graph, is a radical departure from the traditional, tree-based approach of the past, and facilitates rapid and efficient browsing of the document’s hierarchy.ududThis thesis discusses the various aspects of the TechRead system, which will ultimately provide increased accessibility for blind people to technical documents. We demonstrate how the methods used in TechRead differ from those previously employed to solve this problem.