3 ms·
Doesn't speech recognition require either massive databases or massively-parallel processing (or both?) In most cases the speech would be used when connected t
by ck2 17y ago
Doesn't speech recognition require either massive databases or massively-parallel processing (or both?)
In most cases the speech would be used when connected to another server anyway, so it would be okay then to rely on remote processing. Unless I am missing a concept here.
- jacquesm 17y agoIt's funny how natural you take to the remote processing of speech, when I first outlined that as a concept to potential investors about a decade ago I was literally laughed out of the room :) But yes, that's definitely an option, but 'limited domain' speech recognition on mobiles is a fact today, to do better would require a lot of work but I think that it is possible. Don't forget that you'd be about as 'personal' with that computer as possible, nobody else would ever use it, so one dimension of the problem can be collapsed to '1'.