What blue bear said, but in a nutshell, if you consider that a sound source will have a varying volume level, it's designed to reduce the volume of the higher volume sections in such a way that you can't hear the difference in the sound quality, just the volume.
It applies a compression ratio - meaning that for every 1unit you go over a threshhold, you compress the volume by 2:1, 3:1 etc, up to infinity:1 (limiting) - you also have control over how quickly it acts, how hard it acts, how quickly it releases etc.
But read the big article and you'll understand more