Translating a highly context dependent visual language into a slightly less context dependent spoken/text language with different grammar... Also, supporting a

Translating a highly context dependent visual language into a slightly less context dependent spoken/text language with different grammar... Also, supporting all the different sign languages and their spoken counterparts since the market is small enough that you better go international early, unless you want your product cost (close to) five digits a piece like most accessibility hardware. You have more problems to solve than "just" isolating the signs (which the next generation kinect probably could help with) and applying labels to them, but it's certainly a worthwhile challenge, both in effort and in result.