Well, our scientific analysis shows that the voiceprint doesn't match up 100%.
The voiceprint is regenerated responsive to a change to the trainable universal background model.
