Communications of the ACM
Where Siri Has Trouble Hearing, a Crowd of Humans Could Help
A new crowdsourcing program could enhance voice-recognition applications like the iPhone's Siri.
Credit: Beck's Badinage
University of Rochester researchers have developed Scribe, a rapid-fire crowdsourcing program that they say could provide new ways to enhance voice-recognition applications.
Scribe's algorithms direct human workers to type out fragments of what they hear in a speech. The program can direct workers to unique but overlapping sections of a speech and then give them a few seconds to recover before asking them to type again.
Scribe uses natural-language processing algorithms to string together the typed-out fragments into a complete transcript, and can produce a transcript of captions with a delay as short as three seconds using as few as three words. The researchers tested Scribe with workers from Amazon's Mechanical Turk.
The crowdsourced work appears to be only slightly less accurate than that of a professional stenographer, according to Rochester professor Jeffrey Bigham.
"What Scribe is starting to show is the ability to work together as part of a crowd to do very difficult performance tasks better than a person can do alone," Bigham says.
From Technology Review
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found