ID R&D Solution Can Identify a Voice with 1-2 Seconds of Natural Speech

“What’s more, the system is highly accurate, offering an Equal Error Rate of 0.3 percent on microphones and 0.6 percent for speech collected through a telephone.”

ID R&D has upgraded IDVoice, its voice recognition system.

ID R&D Solution Can Identify a Voice with 1-2 Seconds of Natural Speech

The company says that IDVoice can now identify a speaker within one to two seconds of natural speech, making it considerably faster than the seven- to 10-seconds that ID R&D says is the industry standard. What’s more, the system is highly accurate, offering an Equal Error Rate of 0.3 percent on microphones and 0.6 percent for speech collected through a telephone.

In a statement announcing the upgrade, ID R&D’s Chief Science Officer, Konstantin Simonchik, asserted that thanks to its “groundbreaking algorithms comprised of x-vectors and Convolutional Neural Networks, IDVoice offers the most accurate, fastest, and easiest solution for all industries looking to enhance security and UX through voice biometrics.” Simonchik added that IDVoice represents “the next step” in ID R&D’s effort to offer “AI-based multimodal biometrics solutions that deliver unmatched security through a wholly frictionless experience for the end user.”

Other solutions offered by the company include IDBehave, a lean behavioral biometrics system, and IDSquared, a mobile login system combining facial and behavioral biometrics, among others. The company’s technology was recently integrated into Samsung SDS’s Digital Identity Platform.

July 12, 2018 – by Alex Perala