Archimago has a new DAC listening test

[DSJR]

When my ears were only twenty one years old or so and I was doing 'turntable comparisons' using the same stylus each time swapped from one deck to another, any perceived differences I felt I heard disappeared after two or three goes back and forth and after half an hour, I had to go and do something else for my ears and faculties to recover...

Moving forward forty five years or more, I'd urge anyone doing these comparisons to not continually swap between them in a prolonged session, but to try in a relaxed way and not concentrate too hard. If there's any difference at all, it'll creep up on you I feel...

 

[kemmler3D]

The test has nothing to do with sinad,

My bad, I should learn to read.

 

[melomane13]

i am sure than they sound the same, but the methodology is not ok: if the dac with which I listen is less competitive than those tested, how could I hear any differences?

 

[tmtomh]

i am sure than they sound the same, but the methodology is not ok: if the dac with which I listen is less competitive than those tested, how could I hear any differences?

You raise a good question, and to my eyes it's similar to what @Brian Hall commented earlier as well:

Is this really a valid test? Aren't we just hearing our own DACs version of what was recorded?

I think it's about audible transparency: if your current DAC converts digital to analogue with levels of noise, distortion, and frequency linearity that are equal to or better than human hearing thresholds, then your DAC should be good enough for you to be able to detect any audible differences in the output of the three DACs in Archimago's test.

Or to put it another way, if you believe that different DACs can be excellent performers when it comes to distortion and noise and linearity but still sound different in subtle ways then your DAC might impose its own "sound signature" on all three of Archimago's samples, but that doesn't mean your DAC's "sound signature" will obscure or obliterate the differences between the three DACs he used to generate his sample files. To use a crude visual analogy, say the three DACs' outputs are all very "clear," but one's output is slightly "blue," the other's is slightly "red" and the other's is slightly "green." And say your own DAC is also very "clear" but slightly "green." Your DAC's "green tint" will change how you perceive the color of the three DACs Archimago used in his test: the red one will look brownish, the blue one will look blue-greenish, and the green one will look extra green or darker green. But it won't change the fact that the three DACs produce perceptually different colors. So your own DAC could alter your preference among Arch's Samples A, B, and C, but that doesn't mean it will destroy your ability to detect a difference.

Now, in my opinion none of this matters because chances are that all three of the DACs Arch used, plus the DACs used by 99 to 100% of us who are participating in this listening test, are all audibly transparent. Not to mention, I know he used a very nice ADC to capture the analogue output of those DACs, but in general if there's a weak link in a DAC-ADC-DAC chain of processing in terms of measured specs of noise and distortion, it's usually the ADC anyway, not the DACs. But again, a very good quality ADC is IMHO also going to be audible transparent.

 

[Sokel]

If someone present an successful ABX between those files where he will show he can choose amongst them it would mean that he can either distinguish a difference between -120 and -160(!) db attenuation at 22Khz and about 1db difference in FR from 15Khz and upwards or the differences in phase. (The later (-160db) must have a really nice filter admittedly)

I would REALLY like to see that.

 

[DVDdoug]

Moving forward forty five years or more, I'd urge anyone doing these comparisons to not continually swap between them in a prolonged session, but to try in a relaxed way and not concentrate too hard. If there's any difference at all, it'll creep up on you I feel...

That's fine as long as the tests are blind, level-matched, and the results statistically valid/repeatable. There are no time limits in an ABX test but the people who do this kind of thing routinely, find that our short-term hearing memory is better so it's usually easier to reliably hear a difference with fast switching.

 

[melomane13]

You raise a good question, and to my eyes it's similar to what @Brian Hall commented earlier as well:



I think it's about audible transparency: if your current DAC converts digital to analogue with levels of noise, distortion, and frequency linearity that are equal to or better than human hearing thresholds, then your DAC should be good enough for you to be able to detect any audible differences in the output of the three DACs in Archimago's test.

Or to put it another way, if you believe that different DACs can be excellent performers when it comes to distortion and noise and linearity but still sound different in subtle ways then your DAC might impose its own "sound signature" on all three of Archimago's samples, but that doesn't mean your DAC's "sound signature" will obscure or obliterate the differences between the three DACs he used to generate his sample files. To use a crude visual analogy, say the three DACs' outputs are all very "clear," but one's output is slightly "blue," the other's is slightly "red" and the other's is slightly "green." And say your own DAC is also very "clear" but slightly "green." Your DAC's "green tint" will change how you perceive the color of the three DACs Archimago used in his test: the red one will look brownish, the blue one will look blue-greenish, and the green one will look extra green or darker green. But it won't change the fact that the three DACs produce perceptually different colors. So your own DAC could alter your preference among Arch's Samples A, B, and C, but that doesn't mean it will destroy your ability to detect a difference.

Now, in my opinion none of this matters because chances are that all three of the DACs Arch used, plus the DACs used by 99 to 100% of us who are participating in this listening test, are all audibly transparent. Not to mention, I know he used a very nice ADC to capture the analogue output of those DACs, but in general if there's a weak link in a DAC-ADC-DAC chain of processing in terms of measured specs of noise and distortion, it's usually the ADC anyway, not the DACs. But again, a very good quality ADC is IMHO also going to be audible transparent.

human hearing thresholds is the problem.
After some tests i have made and read serious paper on audibility of distortion, it is clear to me that the threshold is really high, -50db thd, more if only present of the second order and third thd.
I can't hear -20 db , 10% H2 on music!
in practice only noise can be a problem, not distortion and not frequency response, since it can be linearized

 

[tmtomh]

human hearing thresholds is the problem.
After some tests i have made and read serious paper on audibility of distortion, it is clear to me that the threshold is really high, -50db thd, more if only present of the second order and third thd.
I can't hear -20 db , 10% H2 on music!
in practice only noise can be a problem, not distortion and not frequency response, since it can be linearized

Agreed.

 

[MaxBuck]

... my guessing results were surprising - not because I guessed wrong, but rather because I guessed wrong every single time. I mean, what are the odds of that?

Based on my success in guessing what my wife really expects me to do, I'd say the odds can be pretty good.