Ha great question Lloyd, I'm not sure there's any point because the original signal got small bits trimmed off it when it was digitized. Dither smooths over that truncation but information is almost always lost unless the signal had a high noise floor to begin with. So all we have to work with in digital is the 16bits coming from the front-end ADC. Digital is always going to be an approximation, calculus assumes the use of real numbers but digital is limited to integers (even when floating point is used the mantissa's still an integer, 24bits for single, 48 for double never a real). We can of course extend the precision as far as we want (or can afford) but integers will never attain the precision achievable with reals, just forever approach closer and closer.
Got it! Thank you...yes, that makes sense. The math applied to the D-to-A would be the same as the A-to-D...so we are ALREADY talking about approximation on the way IN from the original music being recorded, then processed. the way OUT is basically the reverse using the same math. Great to learn!