The Loh Down On Science

Mood Computer

Is your dream date a computer?

This is Sandra Tsing Loh with the Loh Down on Science.

Forget Match dot com. Software engineers at the University of Rochester have developed software that expresses empathy!

The team wanted to develop speech recognition technology that gauges emotions, independent of what's being said. For example, someone says, "I’m happy!" But their tone conveys, “I’m really sad!”

To translate that difference into data, the engineers recorded actors reading neutral text. Things like numbers and calendar days. As they read, the actors changed their voices to convey different emotions. Then the engineers wrote software that matched 12 vocal features--like loudness, pitch and resonance--to each emotion.

Later, when testers spoke into a microphone, if the voice matched the software's happiness data? A smiley face appeared on the monitor! Sadness? Anger? Even pride? A matching emoticon!

Right now, the software works best when the voice speaking is the same one that did the training recording, but the team is working to change that.

Anyway, it could still be a great dinner companion. You know, compared to some.

The Loh Down on Science is produced by LDOS Media Lab, with 89.3 KPCC Pasadena, California. And made possible by the generous support of the Gordon and Betty Moore Foundation.

Follow us on Twitter.


blog comments powered by Disqus