Researchers at the University of Rochester have been working on software that can allow a computer to figure out — with 81 per cent accuracy — how a speaker is feeling based on his or her voice. The recognition is only based on tone, so the actual words don’t matter; if you were to scream at it in Klingon or moan at it in gibberish, it would still be able to get the drift.
The research is still in preliminary stages, but it’s already been baked into a prototype app. So far that app just displays it’s answers, sort of a party trick, but obviously the tech could be wrapped into other functionality down the road. You’d be hard-pressed to make a phone that works better when you yell at it — apart from hamstringing it in the first place — but someone’s bound to find a way to work this kind of tech into carrier bloatware or something. Until themn, you’ll just have to yell at your phone in vain. [University of Rochester via gizmag]