You are assuming these are sample accurate. They are not. They originate from 2 entirely separate analogue sources and therefore could not be. My tolerance in synching these by ear was in the neighborhood of 10 samples - as the Dolby Stereo track had a constant tendency to randomly drift ahead and behind the DC48 track. I was most critical when audio had to exactly match to something in the picture – like lipsynch. If you were to examine waveforms at several different points and find a consistent offset of minus 108 samples with respect to the DC48, then I would say there is a problem. Otherwise, I would consider the coarse offset to be the same for both.