Actually, it's the other way around. On the demo video the presenter says "the audio signal is transmitted from the computer to the phone".

here at 2:15 : http://www.youtube.com/watch?v=X5I6qV2b5iw

I am more curious about if the phone has to be always listening and what happens if there is noise in the environment.

Also what if the attacker manages to bring a recording device close enough? Is the distance important because of the amplitude or because of the delay due to the speed of sound?