Breakthrough in voice biometric optimization comes with IDVoice 2.11

ID R&D announced the newest release of its voice biometric solution, IDVoiceTM version 2.11. This release delivers a breakthrough in voice biometric optimization with an AI-based SDK of under 1MB, enabling voice biometrics to be implemented on the edge - from chips to IoT devices - for a new era of security and personalization in the voice-enabled UX.

IDVoice, ID R&D’s core voice biometric engine, delivers fast, frictionless, and secure voice-based authentication for a range of use cases including mobile and web-based applications, chatbots, IoT devices, call centers, and physical access systems. With the newly reduced footprint, voice biometrics can be embedded in everyday consumer devices like home automation systems and smart TVs.

The newest release of IDVoice includes a number of key performance improvements including:

• Higher accuracy and speed through modified x-vector algorithms: ID R&D’s unique modified x-vector approach differs from those using traditional i-vector by delivering faster and more accurate results when matching voice samples. IDVoice v2.11 builds on the company’s previous x-vector approach to deliver even lower error rates, which led to the company finishing the 2019 NIST Speaker Recognition Evaluation Challenge with exceptionally strong results.
   
• Reduced footprint for wider deployment: ID R&D’s x-vector approach has also enabled the company to decrease the size of its biometric engine from 40Mb to less than 1Mb. Now IoT application developers and device manufacturers can integrate voice biometrics into smaller and less powerful devices which was previously impossible due to memory and processor limitations.
   
• Deployment flexibility: IDVoice v2.11 is available in a number of special fusion versions that can be optimized for specific use cases and hardware, from small IoT devices with limited computational power and memory to large call centers.

IDVoice v2.11 is currently being deployed in various real-world situations, including a mobile banking app, call center fraud detector, and real-time passive call center authentication solution.  

IDVoice, built on ID R&D’s biometric engine architecture, is designed to be independent of channel, language, age variance, as well as to be highly resilient to environmental noise. IDVoice includes diarization that detects and separates multiple voices for accurate authentication in audio streams with more than one person.  

To enable faster adoption and integration by enterprise developers, IDVoice includes wrappers in C++, Python and Java, and ships as a Docker image for flexible cloud deployments. IDVoice supports iOS, Android, Linux, and Windows, making it suitable for mobile devices, servers, private clouds, and embedded IoT architectures. It is available for both text-dependent and text-independent applications.

“With rising customer demand for conversational interfaces, ID R&D is committed to improving deployment, performance, and the user experience, no matter the channel or application,” said ID R&D Chief Science Officer, Konstantin Simonchik. “With IDVoice v2.11, developers can offer advanced voice biometric authentication on even the smallest device, opening up a whole new world of possibilities for securing the connected world.”

In addition to IDVoice, ID R&D offers a suite of technologies and products, including face and behavioral biometric authentication and voice and face anti-spoofing.