You can use Audacity and sample a place in the file where there is just "noise" and no speaking. Then it goes back and removes the noise. It's harder to do in practice because if you remove too much noise it sounds weird. I've tried it before and it takes awhile.
I believe there are actually two recording sets in circulation (of the same exact lectures), or else someone just screwed theirs up somehow and then put it into circulation, because when . . . someone I know . . . obtained the lectures somehow, they apparently obtained two copies, one "normal", and one that sounded pretty crappy.