what would you programmers out there think about starting a self-made voice-recognition project?

would there be a possibility to do something like that?
someone out there who would be able to do that?
any ideas?