Gov’t speaks to voice identification

For a while it was beginning to look like all biometric authentication schemes were going to default to fingerprint technologies. Not that there’s anything wrong with using fingerprints, but there are other methods too. But just as the U.S. government’s Department of Homeland Security has given a major boost to fingerprint methods (outlined, in part, two weeks ago in this space – see the link below) it appears that it is also giving a leg-up to voiceprint technology.

The Operation Safe Commerce Initiative (OSCI) is a pilot project to increase security in U.S. shipping ports and it is being set up by Arlington, Va., security technology specialist System Planning (SPC). One of SPC’s subcontractors is Nova Scotia’s Diaphonic, a company founded by a couple of voice-technology gurus from Aliant, which they describe as “the third largest telecommunications firm in Canada.” (Sounds like the setting for a CBC comedy series, don’t you think?).

SPC will be using Diaphonics’ “Spike Server” (its logo is a bulldog with a spiked collar), which provides authentication and authorization to the initiative.

Users prep by reading a series of standardized phrases to Spike, which abstracts key voice parameters to create the biometric signature. After that, authentication simply involves speaking into a microphone.

For this initiative and demonstration project, Spike will be used to guard access to containers at the port. Anyone wishing to seal or unseal a container will first need to be authenticated by voiceprint and only when the authentication succeeds will they be allowed access to the container itself. It’s simple and straightforward, but it should also be effective.

One question often asked about voice authentication refers to the change in a person’s voice when they have a cold. According to Diaphonics, though, a cold doesn’t affect the error rate of their system although severe trouble, such as laryngitis, could be a problem (i.e., if you can’t be heard you can’t be verified).

I haven’t spent a lot of time with voice systems, but there are a number of uses that immediately comes to mind. Coupling voice with a proximity card, for example, could significantly improve the security of automatic door systems. Or, perhaps, combine voice authentication with voice recognition for PCs. Speaking the name of a file or application not only calls it up on screen, but also provides the authentication and authorization to use it. That would appear to be even better than a lot of single sign-on methods.

Perhaps I’m jumping too far ahead here since I don’t yet have any good numbers on reliability but I’m sure those of you with voiceprint experience will let me know how feasible (or not) these things are.