Huh, thanks for investigating more closely. I'll have a look at that loop. It might be a system-specific issue (at least in that certain hardware might emphasize different aspects of the sample) since I think I would've heard that by now, but I can try to tweak it.
Maybe if I turn the volume way up over here it will become more apparent. Thanks!
Edit: On closer inspection, it is an 816ms sample, by the way, which puts its length close to what you're talking about and that explanation does make sense. It's probably just a case of making sure the loop transitions better.