Thanks to camsr, kraku, purplesunray, music engineer and others for the ideas.antto wrote: maybe this will help:
the sample points (which many audio programs visualise as dots) are not equal to the amplitude
they are more like weight coefficients for the resampling filter
these dots happen to "fall" onto the actual continuous signal only at the very low frequencies, while the more you get towards Fs/2 - the dots go wild, but that's actually fine
Antto, am guessing you wrote slightly different than your understanding. Or maybe it is my misunderstanding. So far as I know which is not much, a properly bandlimited signal will intersect the discrete sample dots even at high frequencies.
For instance that good video which kraku posted, somewhere in it the fella shows bandlimited computer display of high frequency sine wave, where the interpolated line always passes thru the dots. However, at high frequencies, the data points are so sparse that it would take an expert well-trained eye to imagine the proper bandlimited signal shape which would intersect that "apparent random scattershot" of dots.
Frequency vs Time domain are two sides of the same thing, but I grew up with analog scopes and probably think in time domain more often than freq domain.
One can make an analog sine wave sweep, perhaps 10 kHz to 20 kHz, which varies in amplitude according to instantaneous frequency. And one can make an analog sine sweep with solid steady amplitude at all frequencies. With an analog scope, it is easy to see the difference between the two. The solid-amplitude sweep will stay the same amplitude all the way up on the analog scope, and the variable-amplitude sweep will vary in amplitude on the scope display.
I never owned a digital scope. I suppose the spensive ones would have just as true display as analog, but wonder how good the $500-$1000 digital scopes would do on the above task? Maybe fabulous, or maybe stairstep-city? Dunno.
So anyway, a computer graphic interpolation which is good enough to "connect the dots" with plausible-looking curves, but shows amplitude squiggles even on a steady-amplitude high-freq sweep-- Because it always shows squiggles, there is no way to know whether the analyzed signal is steady or has its own built-in amplitude squiggles.
Many digital filters can have built-in fine-grained high frequency amplitude squiggles. A couple of years ago, that is why I noticed Cool Edit Pro was "a little weird" for close examination of high frequencies. I was testing filters with slow sine sweeps. CEP was my "most accurate" graphic display but it wasn't accurate enough to know which HF amplitude oddities are filter characteristics versus which HF amplitude oddities are mere program display artifacts.
Such tiny display oddities would make no difference for 99.9 percent of music-making tasks. You could generally make music just fine even with a crude stairstep or connect-the-dots-with-straight-lines display. Just like, for most purposes you don't need an oscilloscope in the studio to make music. Though I'd guess tasks such as mastering or audio restoration could benefit from computer display which performs near-identical to analog memory scope.
Long ago had a 1970's era giant Tek 100 MHz 4 bay lab scope with "true analog" memory, rather than the later digital memory. But for most tasks digital memory was lots better. The analog memory feature was the first to fail on that old scope, but the thang was so rugged it took years to die. Just lose yet another feature every year or two. It had so many features to begin with, was a long lingering demise. Until finally so much failed that there wasn't much that it could do anymore.