I've never tried it so I dont kmow, I stole the idea from someone who did this to build a solar tracker and it worked well, servo's tend to be very accurate so I'm inclined to say it will work well.
The drawback however would be the light level, the feedback pot in a servo is probably about 10k, the ambient light on the photocells would need to be bright enough to be similar, the input impedance of the servo chip will start to dominate and mess things up under a certain light level.
Easiest way to do it disceretly is 2 mic's, the mic triggers a 555 for 0.1 sec or so, one 555 triggers a h bridge or relay for the motor to go one way, and the other vice versa, nad you need to reset one 555 from the other so they both dont trigger at the same time and short the supply.