Third-party test shows ID R&D leap forward in voice biometrics accuracy
ID R&D has announced major gains in the accuracy of its voice biometrics, with a 0.01 percent false acceptance rate (FAR) at 5 percent false rejection rate (FRR) for device unlocking through biometric authentication in third-party testing.
The company says that up until now, the voice modality could not meet the security standard for mobile device or laptop unlocking, relegating voice to the position of a useful convenience for a limited range of applications. The increased accuracy level, however, now rivals a PIN, according to ID R&D, opening up new practical applications for voice.
Enabling voice biometric authentication for device unlocking, perhaps along with a wake word in the style of voice assistants, gives users the ability to carry out hands-free logins. Voice could then also be used to mobile and web applications.
The advance is described, along with the possible applications it can power, in a new white paper from ID R&D titled “Voice Biometric Revolution: Why Voice ID Is Now Secure Enough for Device Unlock”. In a test for authentication with a wake word using text-dependent speaker recognition, a random command using text-independent speaker recognition and anti-spoofing for both, the third-party evaluation an FAR of 1 in 50,000 in indoor and driving environments, with an average FRR of 9.9 percent. This meets the Android Compatibility Definition Document (CDD) threshold.
“ID R&D is laser-focused on R&D efforts that move the market forward and enable new and exciting use cases for biometrics,” says ID R&D Chief Scientific Officer Konstantin Simonchik. “Our modern voice biometric algorithms consistently push the limits of what’s possible on voice-enabled devices in terms of size, speed, performance, and convenience. As voice becomes the de facto standard for interacting with everything from our televisions to our cars, biometrics emerge as the most convenient way to quickly identify users for security and personalization.”
The previous generation of ID R&D’s voice recognition has scored well for biometric and anti-spoofing accuracy in several benchmark tests and competitions, including the NIST Speaker Recognition Evaluation CTS challenge and the Global ASVspoof Challenge.
The latest accuracy upgrade follows a series of announcements from the company as it makes its voice technology more widely available through availability on the Microsoft Azure Marketplace and WhatsApp for Business, and a partnership with Synaptics to bring voice biometrics to the network edge for IoT and smart home devices.