Voice Authentication

The “Bastard Child” of Biometrics

I am the queen of “zeroing out.” This is call center lingo for those persistently unwilling customers who refuse to cooperate with call center IVR (Interactive Voice Response) systems. You know those absurd 800 menu selections from hell that run you in endless loops which — after several minutes of frustrating button-pushing — deliver you back where you started. Personally, I always figure out how to reach a human operator , even if I have to listen to elevator music while on hold. Hey, that’s what speakerphones are for!

Circumventing IVRs
Recently, however, these so-called “customer service” systems have made it more difficult to reach a human. Often when you push “zero” you are informed that it is an invalid selection. You really have to get clever and strategically navigate to a menu level where you can press a “zero” or at least where you know the unmentioned “five” button option will get you through to a live person. Unfortunately, by the time I do get through to a live person, more often than not I have to repeat all the numbers I had originally punched in when the system answered my call. AARGH! Am I ready for voice authentication or what?

Infrastructure Anomaly
Voice biometrics is unique within the industry in that deployment of the technology does not face the infrastructure challenge that other biometrics must address. In industrialized nations, telephones are ubiquitous. In the United States, for example, there are more then 250 million telephones (as opposed to only 25 million PCs). And while voice faces the same large-scale solutions issues as other biometrics — enrollment, database management, privacy, spoofing, etc. — the infrastructure edge is a distinct advantage. In addition, voice relies on an extension of the most natural form of human interface — speech.

A Real Biometric?
The biggest hurdle voice may have to overcome is the perception that because it is a “behavioral” metric, it is somehow not a “real’ biometric, thus the “bastard child” designation. In fact, recent test results show that voice is accurate and reliable, and in some cases even more so, than other more “traditional” physiological biometrics . Voice currently represents less than two percent of overall biometrics revenues and is not expected to exceed more than eight percent by 2005. But these projections may be ill-conceived.

Voice authentication will likely prove viable in areas where other biometrics face substantial infrastructure and interface issues. In fact, the best opportunities for voice may be in markets as yet untapped and not even on the radar screen of other biometric technologies.

C. Maxine Most, August 2002