Virtual Worlds have virtually exploded in popularity and have experienced significant commercial success. People from all over the globe are experimenting with the various features that virtual worlds provide. It is deemed to be the future of social networking, online advertising, distant education and potentially surfing the web may closely resemble the immersive experience of virtual worlds. But going forward, reckoning its scope, it has brought back the issues of accessibility and needs attention. Individuals with visual impairments usually use screen readers to access web. Virtual Worlds like Second Life lack textual representation and are not accessible to screen reader users. Since the Second Life like worlds are completely owned and created by their users, most of their content lacks meta information. This becomes a huge barrier to a text-extraction based approach of making virtual worlds accessible to the visually impaired who can benefit immensely from this platform.

This thesis presents the design of a framework which can be used for making virtual worlds accessible to the visually impaired individuals. It provides a command based text interface to Second Life which when plugged in to a screen reader, enables the individual with a visual impairment to access Second Life. We then propose a technique which can feed meaningful meta data to the objects in Second Life, which can then be retrieved as text and processed to make sense to the screen reader users.