Well, here is one Biometric technology I have not spoke of in awhile. Voice recognition is one of those technologies which are out there, it doesn’t have the limelight like fingerprint recognition or vein pattern recognition, but it does have its very specialized market niches, and is doing quite well in them. Some of these niches include financial transactions on the phone and confirming the identity of people who call into the many and thousands of call centers worldwide.
The technology surrounding the use of voice recognition has not changed too much, but that all could change very soon. Researchers at the University of Colorado have possibly discovered a new enhancement to voice recognition. The details: “In Johnson’s new system, users set up their accounts by recording a large number of words and phrases which are sent in an encrypted form to create a template for verification. A major concern with the storage of any biometric template is that template falling into the wrong hands and Johnson and his team have come up with a solution they believe eliminates this risk, called Vaulted Voice Verification . . Vaulted Voice Verification is a system in which voice data isn’t transmitted at all. Instead, software sends two encrypted versions of selected words to the mobile phone. One is the user’s voice and the other is spoken by an entirely different person. Software on the phone then has the job of comparing the two samples and determining which is authentic. It then sends back the answer, rather than transmitting the voice recording, making it susceptible to interception and hacking.” (SOURCE: http://www.biometricupdate.com/201212/university-of-colorado-researchers-develop-new-approach-for-voice-biometrics/).
Interesting enough, this new development actually differs from the existing technology in two very different aspects:
*Large amounts of text have to be recited: This is so in this new solution. But in traditional voice recognition systems, only small phrases or certain words have to recited in order to create a rather decent template;
*A version of encryption is being used: No need to rehash it, the details can be seen be above. This is the first voice recognition solution I have come across which theoretically fortifies the template.
Now you may notice I have used the word “theoretical” in the last sentence. You may be wondering why. Well, first in my view, this is still all in the early stages of research and development, so I don’t think that anything is going to break out into the market place any time soon.
Second, given the very limited amounts of detail which was in the news wire, it claims that the voice sample is never really sent over for processing into a template, rather the software compares the recorded voice sample of the end user and another random voice sample. The two are then compared, to see which is authentic. I guess I don’t quite get that yet, but I give the benefit of the doubt because of the lack of technical information which was present. I mean if a template has to be created, in basic layman’s terms, it has to go somewhere in order for it to be processed, right?
It should be noted here, that with voice recognition, no biological or physiological images are used. Rather, as mentioned up above, it is the voice which is recorded, and then converted into a statistical profile. The most common stats used in this regard are Hidden Markov Models.
Well, I will keep my eye on this new development, and if anything, my hats off to these researches for trying to come up with a way to further fortify the voice recognition templates. Perhaps, they should take a look at my recently published Biocryptography articles???