This kind of Ultrasonic tracking i.e listening for a set of specific frequencies that are too high for most humans to perceive is much less complex then actual speech recognition, and consequently much more pragmatic.