Voice Recognition

Voice dynamics relies on the production of a "voice template" that is used to compare with a spoken phrase. A speaker must repeat a set phrase several times as the system builds the template, just like your cellular telephone.

This biometric technique relies on the behavior of the subject rather than the physical characteristics of the voice and is therefore prone to inaccuracy.

It is an up and coming technology, it is currently easily defeated.  We do not see this as a viable system until at least 2012.  Processor speed must be dramatically enhanced, as must the ability of the microphone to discern the actual voice.  Plus the software must be made more forgiving for tired or stressed words and backround noise. 

As you now realize, it's not here, yet, so don't buy in to “vaporware”.  IBM has been working on it since 1980 and as of 2006, it's not working yet.