AudioBrowser - A web accessibility tool for blind and visually impaired user

Entidade Financiadora e Programa: FCT

Data de inicio: 2002-09-30

Término: 2003-09-30

Abstract:

Blind people are at a severe risk of being left out of the Information Society. Information extraction of meaningful information is a hard task. Software for reading web pages is available in the form of talking browsers and screen readers. A Portuguese talking browser is the final product of this project.The application aims at minimizing the scanning time of a web page thus enabling a talking browser to read a web site at a speed comparable to that of a sighted user.

The starting point of this project is a prototype of a web browser designed for blind and visually impaired users. While the browser has received a very warm welcome by the local community the team strongly believes that further work can greatly enhance the prototype by:

• Through extensive user testing to identify interface flaws, and gain a greater understanding of the difficulties involved in browsing highly graphical web pages by the target population;
• Research on information extraction techniques and heuristics to minimize the scanning time for blind users;

In parallel the team will focus its research on document analysis, and information extraction techniques. Further research on HTML structure to define information views is also planned. The main objective of this research is to minimize the scanning time of a web page, allowing a web site to be read within a reasonable amount of time, thus enabling blind users to use the web with an order of speed comparable to that of a sighted user.

Previously identified missing features will also be added to the browser. These include search mechanisms within the page, and special bookmarking where not only the location of a web page is stored but also the current reading position, enabling the browser to start reading a web page from it left on a previous visit. Standard browser features such as history of locations visited, and favorites will be incorporated in the application.

The web browser will produce speech using a specially developed synthesizer, based on the previous development. This synthesizer will allow control of main voice characteristics as well as prosody features. The speech synthesis module is Windows based and supports SAPI. The development will include refinement of the segmental database and improvement of the prosodic modeling in the aspects of intonation, segmental dutrations and intensity. The general programming structure of the module will be improved and optimized as well.

Elementos do Grupo Envolvidos

José Francisco Creissac Freitas de Campos - Investigador

Diamantino Freitas - FEUP - Investigador (Texto-to-Speech Portuguese Library)

José João Almeida - UM DI - Investigador