In this exploratory project, he will develop a prototype client for Second Life that offers a basic level of accessibility, and which will allow him to assess the feasibility of and technical requirements for a client that is fully accessible to blind players. The prototype client will initially allow blind players to navigate the environment using voice commands alone; it will then be enhanced and extended, as time and resources allow, so as to enable these players to interact in meaningful ways with other players.
That's interesting. Most audio games use keyboard navigation. I don't understand why voice commands are preferred, and why they're developed during the initial stages of the prototype when it would seem to me that the first thing you need is feedback from the world (i.e., spatial audio cues) before you start to move around in it.
Achieving these objectives is not straightforward, because the client and server of Second Life have only recently been made open source and no one has yet attempted to create an accessible client for the environment.
I didn't think the server was open sourced yet, though it is apparently planned for some as-yet unspecified point in the future. I have heard that some people have reversed engineered the network traffic (or merely extracted it from the client source) and have extrapolated their own server based on how it appears to work. The official line from Linden is,
What source code won't you be releasing?
We don't (yet) plan to release the code that runs our simulators or other server code ("the Grid"). We're keeping an open mind about the possibility of opening more of the Second Life Grid; the level of success we have with open sourcing our viewer will direct the speed and extent of further moves in this arena.
There's an interview with Eelke, for further reading too.