I think some big money needs to go into subvocal recognition (if it isn't being heavily invested in already). Devices like these will be a lot more attractive if they can be used without speaking out aloud.