Patents Granted for ArmorVox ImpostorMaps and Cloud Based Identity Authentication ApplicationsJune 20, 2012
Key innovations at the heart of Auraya’s ArmorVox Speaker Identity System technology are now protected under patents granted in Australia and pending in other jurisdictions.
Specifically, the last 12 months has seen Auraya granted two patents, one relating to the applications of ArmorVox for Cloud based identity authentication and the other relating to ImpostorMaps™ a process pioneered by Auraya that enables ArmorVox to automatically set thresholds according to the security performance of the voiceprints enrolled in the system.
Until now, setting thresholds in voice biometrics systems has been a “hit-and-miss” affair.
Traditionally, solutions providers guess a threshold setting and then try out the system and see how they go! Simply putting a security solution out there and seeing how many hackers break-in is not a methodology usually recommended by the IT Security Departments of most major banks, Governments and telecommunications carriers for obvious reasons!
However, with Auraya’s patented ImpostorMap™ technology, which is a built into the ArmorVox product as a standard feature, IT Security Departments now have a systematic approach to setting the thresholds in a voice biometric system and determining the security performance of the solution before deploying into a commercial and potentially dangerous commercial application.”
About ArmorVox ImpostorMap™ Tuning
ArmorVox ImpostorMap™ Tuning is a standard feature of the ArmorVox product. This allows partners to automatically tune and optimize ArmorVox for the language, accent and dialects spoken by the population of enrolled speakers.
Enhanced accuracy for each and every user enrolled on the system irrespective of language, accent or dialect
ArmorVox maximizes authentication accuracy for each user enrolled in the system, tailoring thresholds to the security performance of each voiceprint in the system. This is an important distinction compared to the less precise method of tuning a voice database for the overall population. Our experience conducting “Ethical hacking” engagements shows that the majority of errors made by a voice biometric system are attributed to a minority of weak and vulnerable voiceprints. Maximizing performance on the overall system does not necessarily addresses the weakness and vulnerabilities of individual voiceprints. By focusing on maximizing accuracy for each speaker and – the performance of the overall system can be dramatically improved.
For further information, please contact Auraya.