The University of Michigan and Adobe have developed a prototype of a new interface and application to make image corretions and manipulations. By speech and gesture recognition the application understands what the user wants. The possibility to use areas like top or bottom in the speech commands seems really cool to me.