Voice biometrics vie for role in mobile ID tech
- By John Breeden II
- Dec 16, 2013
The government requires two-factor authentication to access most networks. One of those methods is almost always a password. The other can be anything, including a token system like a Common Access Card, a fingerprint scan, or some other form of password.
Although voice authentication has also been used for security on some systems, it's normally only deployed on large enterprise setups. Now AGNITiO, a voice biometrics company headquartered in Madrid, Spain, wants to change that and move full biometric voice authentication to smartphones and other mobile devices.
Called Voice iD, the software can identify a user's unique voice pattern regardless of the language spoken or the phrases read. The company likens it to a fingerprint, where each person's voice is completely unique. Even if someone is trying to mimic another person, AGNITiO officials say their program won’t be tricked.
Given that a voice is something that a user always carries with them, and the fact that they don't have to remember any type of password, the company sees it as an easy solution to adding a second password authentication method to any phone or network.
“With Apple’s introduction of the Touch ID feature in the iPhone 5S, consumers have awakened to new ways of authentication,” said Emilio Martinez, AGNITiO’s CEO. “Voice iD may be the easiest method, complementary to Touch ID, yet faster to deploy.”
The AGNITiO Voice iD software engine can be added to any device, platform or application, which allows for rapid adoption, the company says, “Soon consumers in a variety of scenarios will identify themselves with their unique voiceprint, in any language, electronically or by phone in a variety of scenarios.”
To combat many of the ways voices can be stolen, such as the method used to defeat a security system in the movie “Sneakers,” AGNITiO has developed an anti-spoofing system. The company says its security is 97 percent effective in preventing taped replay attacks, and 99.9 percent effective against someone simply trying to mimic another speaker.
John Breeden II is a freelance technology writer for GCN.