[Jack Simth]

The reciever just has to have high-fidelity detection; it doesn't necessarily need a recieving mirror at all - a solar panel shaded against direct sunlight feeding a mic input would do the job. From the description below, one simple idea would be that the vibrations on the mirror caused by sound waves would slightly change the angle of reflection, resulting in a measureable difference in reciept at the target; a small target either has the signal or has the signal interrupted - the length of those interruptions then determines the wavelength of the sound wave that produced it. A basic amplitude modulation system, really. A sufficiently large target can actually recieve the beam all the time, but at different locations - in which case, the sound wave gets graphed in it's entierety and you get better fidelity.

Quite frankly though, a laser link is more reliable (not weather-dependant, don't need to constantly adjust for the position of the sun, can use a specfic set of wavelengths to help filter interference from other sources {can even pick frequencies the sun hardly uses so that you can broadcast even with the sun directly behind you}, et cetera).