I am not sure I agree with the basic premise that CD playback is free of any timing error. The digital encoding may be, but during playback, you have a rapidly spinning disc being scanned by a moving laser beam. Those two mechanical devices have to operate perfectly and in concert with each other, if the timing is to be "perfect". Since we live in the real world, I would bet that doesn't happen. Second, you neglect the fact that a human engineer was involved in generating the two different forms of software from presumably a tape or a digital master. Said engineer may well have fiddled with the content such that the two versions of the same piece would be different in time duration from one another. Finally, I don't believe we can reliably detect a time interval as short as 100msec. Did you test yourself for your reaction time? Did you repeat the experiment at least 3 times? That would help to compensate for errors at the start point, if you averaged several results. And finally, if you're going to check timing in this fashion, why not in fact use a strobe, as Chayro suggests?