January 16, 2014 -
Voice biometrics is a rapidly-growing biometric modality. Thanks, in large part, to the proliferation of smartphones and other mobile devices with microphones, but also to changing attitudes towards security and authentication.
Alexey Khitrov, the president of SpeechPro, the American subsidiary of the Speech Technology Center, tells me that this industry has long been “the market of tomorrow,” but it’s now becoming the market of today.
“We’re seeing an unprecedented rate of adoption. We’re moving from early-adopters into a much bigger and wider market,” Khitrov said.
According to recently-published research, the overall market for face and voice biometrics could reach $3 billion by 2018.
The company has been around for more than 23 years, having been founded by a team of engineers and scientists in 1990. SpeechPro has its roots in forensics and has a heavy focus on R&D. Of the 400 people working at the company to date, nearly 150 of them are solely focused on R&D efforts.
Khitrov has been at the helm of SpeechPro for six years and has been leading operations in NYC for the last three. Before that, Khitrov worked for BMO Capital Markets in Canada.
With a traditional focus on B2G solutions, SpeechPro is now seeing great increases in interest from both B2B and B2C sectors and is working to address these new markets, but will also continuing to compete in the B2G space.
As a part of this effort, Khitrov and his team have developed an app – VoiceKey.OnePass – that combines both voice biometrics and facial recognition. Khitrov demonstrated the app for this interview, showing the process of enrollment, and then text-dependent authentication.
Here’s how it works: A user repeats a randomly-generated string of numbers three times and takes a facial photo to enroll. Then, the user needs to recite the numbers 0-9 so that the app knows what they sound like saying numbers other than those in the random password.
After that process – which Khitrov is quick to point out takes less than a minute – users need to line their face up in a template on the screen, and recite a string of numbers for authentication. To ensure there is an actual person reading the numbers and not just a picture in front of the camera, the app takes multiple photos as a form of liveness detection.
“Threshold for acceptance is an art, it’s not a science,” Khitrov said, explaining the customizability of his app. “We work with customers specifically to determine what that needs to be.”
SpeechPro has developed this app internally, but facial recognition component comes from Cognitec.
As we’ve discussed previously in BiometricUpdate.com, automobiles are a likely target for new biometric integrations, and according to Khitrov, the environment is perfect for voice biometrics, and consumers could soon see new solutions come to market.
Late last year, SpeechPro announced that the Criminal Investigations Unit of the Nepalese Police is now using its Forensic Audio Workstation for its forensic and voice identification needs.
Earlier in the year, SpeechPro and Voxeo inked a partnership to deliver SpeechPro’s VoiceKey Intelligent Voice Authentication solution to customers using Voxeo Interactive Voice Response services.