Thanks for the further info. I had been wondering if perhaps the people having trouble were playing things at a very high volume or were holding the phone very close to the speakers or something (resulting in clipping), but that doesn't seem to be the case here at any rate.

Does your speaker system consist of separate tweeters and sub-woofer? That might result in a different mix of bass and treble output than in the original recording, I'd need to investigate how much that might influence the algorithm if that's the case.

I'm just throwing ideas out at the moment though, so if there's anything else you can think of about your set up that might have an effect I'd be interested to hear it.