What measurement are we missing?
Tale of three DACs. All three measure with better specs that the "experts" tell us we can hear.
Schiit Asgard with old internal AKM DAC. So-so measurements on the DAC
Topping D30pro. Incredible measurements by any standard. I mean WOW.
JDS Atom DAC+. Down to the only "darn good" level of measurements, but still below the theoretical threshold of hearing
All measured here by Amirm. Solid, reliable objective numbers.
But
The Asgard I needed to roll off the entire top end and about 4 dB on 3100
Topping I could hear slight detail above the background on headphones better than the Schiit, but not through the speakers. A hair maybe better at 3100. The evil "glare" baby scream frequency.
Atom clearly smoother midrange allowing bringing up the highs about half way restoring the detail I lost taming the other two.
Atom seems to make the worst recordings ( someone give the Billy Joel engineer a dope slap please) only bad. The others seem to accentuate the bad. Great recordings sound great on all three. Maybe a hint to the subjective reviewers. Use bad recordings.
I am no golden ears. I am an old bugger who's hearing now rolls off in the above 16 and not sure goes past 18K. ( when young I did hear to 22). I do know what I hear and my wife who has far better hearing confirms. Not damage, but the normal thickening of the fluid in the inner ear that comes with age. Someday maybe science will have a pill that reverses that. Hard to believe any objective reviewer or mixing engineer older than 50, but I digress.
So, what am I hearing that is different and not captured in the current measurement suite? All vanishing low distortion below the 16 bits of my CDs, ruler flat response, noise way below threshold. All three were routed through the Asgard analog amp section. I did not hear a difference Topping direct or through. These differences were great enough not to require precise AB testing. Three or four CDs were enough and reproduceable. Bias? Of course, but my bias was the opposite of what I heard. I expected the D30 to be cleaner. I did not expect to hear any difference with the JDS. I had expected the more expensive better measuring to sound so. I liked the box. I got the JDS more to verify it was in the "good enough" camp and expected it to sound the same and then to use on my desktop. It stays in my Stereo.
Is there some threshold where additive distortions correlate and then bother us? So a .000-something is winding up at the speakers as .1% from all the steps mic to speakers and we can hear that? Do we need super playback to not accentuate all the previous steps? Failing on less than great material, but squeaking by on great?
FWIW, my older DAC was a first generation Wolfson. Better than my CD player, but we are talking '80's tech and distortion in the .01 range as I measured. Far less detail. Discernable noise with headphones. Required even more top end roll off to be listenable. Specs would have shown that and it would land about an inch to the right of the SNAID chart. Curious though, my cheap SMSL Tripath desktop amp is not as shrill as the Topping even with the Grados, but fatigue sets in pretty quick. I am waiting for a review of the new Topping all-in-one.
It seems, to a point, the classic measurements correlate to sound, but at some point, we need something else. If we hear a difference between the "blue" tested units, what is it? On paper they should be indistinguishable. Timing? Group delay? Something measured we are not focusing on, like the HF filters?
Atom is a keeper.