Does Digital Try Too Hard?


Digital glare. A plague of digital sound playback systems. It seems the best comment a CD player or digital source can get is to sound “analog-like.” I’ve gone to great lengths to battle this in my CD-based 2-channel system but it’s never ending. My father, upon hearing my system for the first time (and at loud volumes), said this: “The treble isn’t offensive to my ears.” What a great compliment.

So what does digital do wrong? The tech specs tell us it’s far superior to vinyl or reel to reel. Does it try too hard? Where digital is trying to capture the micro details of complex passages, analog just “rounds it off” and says “good enough,” and it sounds good enough. Or does digital have some other issue in the chain - noise in the DAC chip, high frequency harmonics, or issues with the anti-aliasing filter? Does it have to do with the power supply?

There are studies that show people prefer the sound of vinyl, even if only by a small margin. That doesn’t quite add up when we consider digital’s dominant technical specifications. On paper, digital should win.

So what’s really going on here? Why doesn’t digital knock the socks off vinyl and why does there appear to be some issue with “digital glare” in digital systems.
128x128mkgus
I've heard things you people wouldn't believe. Attack notes on fire off the shoulder of Mozart. I've heard C-notes glitter in the dark with the PPT Gate. All that music will be lost in time, like tears in rain. Time to die.
Roy, where have you been all this time?


Post removed 
@asvjerry 

Don’t quite see how smoothing is going to help all that much with 2 samples per waveform. Or even 4 (for example when samples are taken at pi/4, 3pi/4, 5pi/4, and 7pi/4).

And never mind phase distortion (by which I mean the distortion inherent when subsequent waves are sampled at different places on the x axis). But is that not the underlying insight of Nyquist/Shannon, that frequency information can be recovered to arbitrary precision at the (temporary) expense of phase information? But real-time sampling still leaves us with this ’temporary’ problem, or so it seems to me.
terry9
All you have to do is draw a sine wave. Then make 250 equally spaced marks on the x axis, starting with 0 and ending at 2pi. Use each mark as the step boundary of a step function. Just like elementary calculus.

Now calculate the mean square difference between that step function and the sine wave, and divide by the sine wave area - it’s about 5%. You may infer that 250 samples per waveform delivers about 5% distortion. Now, how many samples per 20KHz waveform?
This argument suggests that you misunderstand the math that underlies digital audio. You might want to view this.

Don't get me wrong - I'm an analog guy. But we can't improve digital if we don't understand how it works, and how it doesn't.

@cleeds 

Thank you for your friendly and helpful response. I'll take a look and we can discuss.