And you're going to have the sum and the difference frequencies to deal with as well if the transmitters aren't perfectly synchronized. And those differences may very well be within the audible range.

If nothing else, now you've created a new type of Active Denial System

Phasing, and sympathetic resonance. With 10+ speakers pointing in constantly random direction, in an environment as changing as a starbucks, I would assume that somewhere something might start vibrating at a frequency well within human/cat/dog hearing. Just pray it isn't one of your fillings.