I read someplace that human speech recognition is around 90%, I think for speech recognition to be considered "useful" it has to be super-human. I agree somethi

I read someplace that human speech recognition is around 90%, I think for speech recognition to be considered "useful" it has to be super-human. I agree something like 99+% is required before it really gets accepted. One problem lots of speech driven systems have too is the lack of discover-ability. There's no interface you can inspect to learn from unless the system gives you a long list of command choices.