Technology licensing company Sensory has launched TrulySecure 2.0, an updated version of its speaker verification and face recognition solution released a year ago. The systems is for mobile, tablets and PCs and enables multimodal biometric authentication.
TrulySecure’s accuracy was upgraded by using convolutional neural networks and deep learning approaches. “The system uses machine learning technologies to create different speech and vision approaches that can be used for speech recognition and authentication,” says Todd Mozer, president and CEO of Sensory. “We licensed those technologies into consumer electronics historically, and over the last year we’ve moved more towards banking, medical and enterprise applications.”
TrulySecure aims for a technology that is accurate as well as easy to use. “Right now we have an accuracy that’s as good as one in a million false accepts, so it’s quite a low error rate. But we haven’t sacrificed ease of use in the process,” Mozer says. “It will work on pretty much any mobile phone that has a microphone and/or a camera…and it’s fast. It authenticates in about a second and the training process is less than 10 seconds.”
To use the AppLock with TrulySecure app, the individual enrolls by saying a passphrase three times while looking at the device’s camera. “After that, it’s looked at my face and it’s heard my voice enough that anytime it wants to authenticate me, it can do it with my face alone or my voice alone – or my face plus my voice, which is what we call TrulySecure mode,” Mozer says. For ease of use, the user merely puts his face in front of the camera – no need to press any buttons – and the voice can be included for added security.
Mozer says TrulySecure is FIDO Alliance certified and compliant. User information is processed and stored solely on the device, not in a company database. For developers who want to integrate advanced biometric authentication into their apps, Sensory is offering SDKs for iOS, Android, Lenox and PC.