[question000]

I honestly would love if we stopped trying to square the circle with speech recognition and just built a phonetic signal based system for computer interaction. Words are overrated, I'd rather whistle at my speakers like I'm R2D2.

[jobigoud]

I was also thinking about this, in the context of silent speech, trying to recognize phonemes directly from muscle activation on the throat. Instead of trying to pattern match an existing language, which is too complicatex/ambiguous, we should construct a language dedicated to that, that is easy to recognize from the EMG. Then we can have the software 'translate' the message.

[yjftsjthsd-h]

I would also happily go for that, but I suspect that most people do not want to learn a new way to communicate just to talk to their computers.