To me, it looks like they took the guys voice, played it backwards about half a second after he said something, and lowered the decibel level of that track of sound. You can accomplish this using a program called Audacity (it's free, look for it on sourceforge).
