I think we often forget that because the scale is not linear, people with a 10K limit still have very useable hearing. The very top note on a piano is just over 4000Hz, so a whole octave above that - 12 extra notes is only taking it up to 8K - so the first harmonic of the highest note is 8K - 16K, another octave up is nearly two feet to the right of the highest piano key. The SM57s people cite as having quite a low cutoff at 15 K can nearly get the second harmonic of the highest piano key. Those people who can hear up to 20K hear something, that's clear and measurable, but does it help the music? I'm not sure. Lets consider the second harmonic - two octaves above that key's fundamental frequency. The top keys are quiet compared to the lower ones, far less musical, and the harmonics are lower in level as they go up, so we're recording 8K at quite a few dB down, then more reduction for the 2 harmonic and so on. Of course we also have the sums and differences, which again are lower in level, so what is right at the top is not that useful - maybe it's even noise instead of music? I don't know as I can't hear it. I don't think though that my appreciation of piano quality is different to what it was when I was 17. I still hear the differences between pianos - so that difference must be far lower in frequency. If you go to an audiologist, many of their test machines cannot even go above 10K - professional audio clinicians who don't have test equipment for frequencies that high?
Playing pink noise and sweeping the cutoff frequency creates the impression that the loss of HF is important. When I do it, and see what my hearing is doing - three quarters of the way across the screen - I can hear that top end hiss changing, but then stopping changing. If you do exactly the same test on a piece of piano music, the difference the sweeping does is much, much less. You knock off the top at 15K can you tell what has gone if you can hear it. How about 14,13,12 and 10? At some point it just starts to get duller and somehow lower in quality. This I think is a better test to do than tones. You can also do it with 80s synth pop because of the popular saw waveforms - Don't you Want me Baby is a great one to try - there's a kind of sizzle up top that does become obvious when you sweep the HF cutoff. Another good one to try is Elton John's Funeral for a friend - the synth part at the beginning stands out, but once the song kicks off with bass, drums, piano and guitar, the top content is not so impacted. Interesting to dig up your historical favourites and try them. Very few old recordings put anything up top to experiment with. Give it a try.