Companies and organizations alike have long been utilizing voice biometrics. You may currently be thinking just what I thought before I fully absorbed this term: voice bio-what? Voice biometrics, a technology comprised of numerical models of characteristics (i.e. sound, pattern, rhythm) found in an individual’s voice as represented by a voiceprint of spoken qualities, enables specific user recognition within IVR systems that involve higher levels of security.
A Plum Voice blog covered the topic just today, reporting that according to the International Journal of Computer Applications, “the way an individual talks is one of those distinctive features that may be used for recognition. Voice, usually considered as a form of behavioral biometric, is in fact a combination of both physiological and behavioral biometrics.”
Voice biometrics acts as a quick, convenient, and most importantly a secure way of remotely determining an individual’s identity. This leads us to the blatant question: why aren’t more organizations taking advantage of and integrating these functionalities into their IVR systems?
Maybe one reason is due to the fact that creating a voiceprint can be very tedious and demanding; the caller must repeat a specific phrase, number sequence, or even read through an entire paragraph of literature several times in order to collect a sufficient amount of data to construct an accurate and secure voiceprint. After this lengthy process, the system mathematically models characteristics of the individual’s voice, but still, the individual may be asked to repeat words, phrases or sentences again before placing an incoming call.
If you’re on the consumer side looking in, this process may seem unnecessarily long and agitating, perhaps even doing a disservice to the original purpose of this process. Companies offering highly secure information telephonically, however, need to balance convenience with safety on a daily basis; something which is far from easily done.
This poses a problem with voice biometrics, whose highly convenient systems are unfortunately not as secure, leading to a high rate of false acceptance incidents which in turn increases the chance of fraud. Just as problematic, if you want a highly secure application, it almost always has low convenience rates – meaning systems could be too secure. You might be wondering, since when did being too secure become a problem? In actuality, systems that are too secure could falsely reject the input of the correct user. This is a tough situation to be caught in; so how do companies balance between these two imperative components?
Something that could really help simplify the process would be applying personal knowledge questions, which would cut down on the crazy repetition which is currently required. Asking personal knowledge questions selected by the user can not only dramatically speed up the process, but also quicken future verification needs. This would ultimately reduce the amount of false acceptance and fraud while focusing on what’s really most important – maintaining a quick and convenient process for the consumer.
Edited by Jamie Epstein