I won't even pretend that I got it, all those links are beyond me.
ok so here is how my non engineer head sees it(feel free to tell me when I'm wrong and make jokes about how stupid I am. I'm going all out on a limb). I hit a cymbal or whatever, and the first sound is the impact instead of the cymbal vibrating from it. the problem would be that first sound and not at all the cymbal vibrating right? so what I should care about is the impact itself where in a small time, the amplitude recorded may have risen pretty fast and pretty high. much faster than a 20khz sine else we wouldn't even care, it would just be usual signal. on an audible level if it's rising much faster than a 20khz signal, I figure I can't hear it right?
but on an electrical level, I guess that if the amp isn't fast enough to follow, it ends up trying to make the waves with an offset in amplitude from failing to rise fast enough. and it will then be making distortions untill it gets back on track. is it something like that or did I get the all idea wrong?
because if it's like that, then worst case scenario, it would take less than half of the fastest sine the amp can make to get back on track. and that's only when the transient or whatever first impulse was both super fast and super loud. else it will be corrected a lot faster.
do we really record a lot of those high amplitude super fast stuff? looking even at highres tracks in audacity that doesn't seem to be the case much.so I suspect microphones to fail a bit on those stuff(or maybe I just don't talk at all about the right problem? ^_^)
also (again tell me if I'm completely on the wrong planet here) wouldn't just listening to 16/44 music remove those transient sound entirely with the low pass filter anyway?
oh and last, I thought that most audio IMD tests were SMPT... something(don't even know what it means ^_^) with 60hz and 7khz. but that wouldn't really tell much about what happens at 20khz right? are IMD from a 20khz wave usually measured?