Well of course music does involve special skill, which, having listened to Hillary Hahn playing the Sibelius violin concerto earlier this evening I could not deny. And I would also say that playing like that is so difficult that it would be churlish to deny her anything that helps, whether it can be justified on theoretical grounds or not. But as far as say the reproduction of music goes, I came to the conclusion many years ago that it was best to just make sure the gear is good enough that you can live with the imperfections and just listen to the music. So for instance, even knowing that a typical tone arm on a record player has a tracking error of up to 2%, thereby contributing a similar amount of actual harmonic distortion in places along its travel, still provides a good listening experience and most people do not notice it. So why pay over the odds for linear tracking. Of course, now that we have digital recordings, that one need no longer be a concern, except to people who like to think that vinyl is somehow better.
For Michael's point, well, pretty well all of our models of how the universe works are approximations, or are likely to turn out to be so with later discoveries. That does not mean that a pendulum is useless for a clock, the results are good enough to be useful. (so long as you stay on this planet!) And no structure is actually really rigid, but that can be a useful approximation for many purposes. For critical things like aeroplane wings, the flex does have to be taken into account.
John