Mark,
I was sure your remark was a joke, but you couldn't tell that from my response.

The classroom experiments you cite belong firmly into the field of 'Illusions of memory'.

Eidetic imagery is a rare phenomenon (even more rare in adults than in children), so rare that some researchers have declared it as nonexistent. The consensus at this time seems to be that it is a different kind of memory (I really should read whether there are data on that from modern brain imaging methods in which methods from physics like NMR are used to find which brain parts are involved in a recollection), a relict from our evolutionary past which normally gets lost quickly when growing up.

I haven't read about research on the auditory system in that respect (except for the echoic memory, which preserves the acoustic input veridically for a few seconds). There might be a long term equivalent to eidetic memor, but I don't know. What you describe, however seems to me still to be within the normal functioning of memory. Our recollection is much better, if we test recognition than free recall. And our recollection is much better if we test something which was deeply (that is emotionally, semantically) processed. So if you make a recognition test on something which was deeply processed (as a music piece usually is) I'd not be surprised if the performance on that test is good even after years. But that's spoken from the armchair and from results in other fields, I don't know any data on the phenomenon you describe.

Wolfgang