Part of the School of Computer Science’s Information Management Group.
The initial eye-tracking studies provided insight into how updates are used by sighted people: which they attend to, and which they ignore; which parts of the new content they view, etc. This understanding has allowed us to propose a set of mappings, suggesting how each type of update might be presented in audio, so that it is easier to use and provides information in as efficient a manner as possible.
To test these mappings, an audio Web browser has been developed. This is based on the Fire Vox extension to Firefox developed by Charles Chen. Fire Vox is a screen reading extension that turns the Firefox Web browser into a self-voicing application. It is notable for its support of WAI-ARIA.
Although heavily based on Fire Vox, the SASWAT browser feels quite different from Fire Vox, using a different user-interface, and handling updates quite differently. Instead of using ARIA tags to determine how to present an update, this system classifies it according to the content and the user’s activity, and bases presentation on the class. This has allowed us to test the rules to determine if they help users.
The development of this system has been targetted only on creating something to test our presentation model, and the software is therefore less than robust. A special evaluation Web site was created, on which the browser behaviour was (generally!) predictable. At present, the system is not suitable for general Web browsing. The testing and evaluation have all been done using the Mac Leopard system voice and Firefox 3.5+.
The user-manual is available from our Lab Repository. The extensions are also available for download here. Four extensions need to be installed; the first three are available from our Lab Repository; the fourth can be downloaded from the Fire Vox site.