AES Paper Digest: Do Audio Op-amps Sound Different?

[Krunok]

Because they can do AB Tests on the computer of captured files which is a lot easier than hardware.

Sure, I'm aware it's easier that way, but I think it's not the way it should be done. In my opinion analog output signal from the DACs should go to the switcher and than to the preamp/amp/speakers chain. As we have seen from your measurements output signal from modern DACs differs very slightly so doing ADC can affect those differences that are anyhow very small.

 

[Krunok]

And doesn’t this reflect the Swedish crowd’s objective bent? They don’t care if the signal was put through an AD stage because electronics are pretty neutral these days. Right?

They don’t believe the AD stage will colour the sound in a way that will mask audible differences if they exist. Agree?

No, I don''t agree. If they believe ADC doesn't affect the way we hear music than neither the DAC does. And if that is the case, than why bother to measure the difference between DACs? Doing ADC after DAC kind of defeats the sole purpose of their measurement.

 

[RayDunzl]

They don’t believe the AD stage will colour the sound in a way that will mask audible differences if they exist. Agree?

Another option is they think the extra AD will color, but color both equally, leaving the convenience and difference factors intact...

 

[Krunok]

Another option is they think the extra AD will color, but color both equally, leaving the convenience and difference factors intact...

Yes, I believe they were counting on that, but nobody proved that really works that way so they shouldn't use that method. In my opinion that only makes differences even harder to hear as they are masked with ADC.

 

[svart-hvitt]

No, I don''t agree. If they believe ADC doesn't affect the way we hear music than neither the DAC does.

I think they implicitly assume that for significant differences to exist, they must be audible through a good AD stage.

Remember, the human perception is not a measuring device. So why not make a more robust test aimed at humans that will only discover significant sound differences?

 

[svart-hvitt]

Another option is they think the extra AD will color, but color both equally, leaving the convenience and difference factors intact...

Agree.

 

[oivavoi]

And doesn’t this reflect the Swedish crowd’s objective bent? They don’t care if the signal was put through an AD stage because electronics are pretty neutral these days. Right?

They don’t believe the AD stage will colour the sound in a way that will mask audible differences if they exist. Agree?

Yes. And the fact that some devices appear as transparent, while others do not, is an indication that their chosen AD stage doesn't color the sound in any noticeable way.

 

[oivavoi]

No, I don''t agree. If they believe ADC doesn't affect the way we hear music than neither the DAC does. And if that is the case, than why bother to measure the difference between DACs? Doing ADC after DAC kind of defeats the sole purpose of their measurement.

I don't think they believe that all ADCs are completely transparent, but I think they chose an ADC for this task which is as close to transparent as it gets, according to their measurements and listening tests.

 

[svart-hvitt]

ROBUSTNESS

If a little room for error is allowed in the AD stage, while you win robustness in the test procedure and method, the chosen method sounds (sic!) sound.

 

[Krunok]

I think they implicitly assume that for significant differences to exist, they must be audible through a good AD stage.

Remember, the human perception is not a measuring device. So why not make a more robust test aimed at humans that will only discover significant sound differences?

I don't see how such such assumption can be justified. Based on what exactly? Making such unproven assumptions is not the way scientific experiment is supposed to be made.

I also don't see how doing ADC instead of using hardware switch made test more "robust"???

 

[Krunok]

ROBUSTNESS

If a little room for error is allowed in the AD stage, while you win robustness in the test procedure and method, the chosen method sounds (sic!) sound.

And how do you know the error is "little"? And why are you saying test is more "robust" with files than with switch?

 

[Krunok]

Yes. And the fact that some devices appear as transparent, while others do not, is an indication that their chosen AD stage doesn't color the sound in any noticeable way.

I totally disagree. You are drawing conclusions based on test results which may be spoiled by ADC. That is not the way to interpret the measurements.

 

[oivavoi]

And how do you know the error is "little"? And why are you saying test is more "robust" with files than with switch?

Point is: If this test is able to generate both positive and negative results (there are devices which go unnoticed and devices where they detect coloration), it implies that the test procedure works somehow. If it was the ADC that was responsible for the detection of some devices, the logical conclusion is that it would work the same way for all devices under test.

 

[svart-hvitt]

And how do you know the error is "little"? And why are you saying test is more "robust" with files than with switch?

I don’t know the error is little. I trust Mr. Granqvist, I guess, who seems to be a competent scientist. See also Ray’s comment above on significance of error in AD stage.

If you rig a test in the digital domain, it will be more robust if analog, manual steps were involved in the test.

 

[Krunok]

Look guys, no device does perfect ADC, which means that unknown amount of noise and distortion has been added to the measured signal, which wouldn't happen if they have been using the hardware switch. So I really see no point in justifying the ADC when it is obvious it affected the measured signals in an unknown way.

 

[RayDunzl]

So I really see no point in justifying the ADC when it is obvious it affected the measured signals in an unknown way.

If they are sending the "test" to various locations/people, then files are easier to send than hardware.

I think somebody here has done repeat-loop ADC->DAC-> ADC->DAC and needed to do eight (?) passes before he could detect (ABX of the files) difference.

The files are/were posted.

---

Here it is:

https://audiosciencereview.com/foru...dac-loop-vs-the-original-can-you-hear-it.448/

 

[Krunok]

Point is: If this test is able to generate both positive and negative results (there are devices which go unnoticed and devices where they detect coloration), it implies that the test procedure works somehow. If it was the ADC that was responsible for the detection of some devices, the logical conclusion is that it would work the same way for all devices under test.

To make such conclusion you should be totally sure that ADC didn't introduce any effects that can be heard. But you can't be sure in that as that has not been proven.

 

[svart-hvitt]

Look guys, no device does perfect ADC, which means that unknown amount of noise and distortion has been added to the measured signal, which wouldn't happen if they have been using the hardware switch. So I really see no point in justifying the ADC when it is obvious it affected the measured signals in an unknown way.

If AD produces noise and distortion way below audibility, I don’t see how the AD stage would make the Swedish method fragile.

However, I do agree that the AD stage may let some DACs or gear go under the radar, in theory, at least.

I guess I’d be willing to live with that risk given the test is otherwise more robust than an analog, manual one.

Very few test procedures are bullet proof. That’s where subjectivism enters the scientist’s world: He has to make choices between imperfect alternatives, i.e. decision making under uncertainty.

 

[Krunok]

If they are sending the "test" to various locations/people, then files are easier to send than hardware.

Sure, it is easier. But I believe tests shouldn't be done the easiest way, but the more correct. way.

I think somebody here has done repeat-loop ADC->DAC-> ADC->DAC and needed to do eight (?) passes before he could detect (ABX of the files) difference.

Yes, I believe that was the case. But in that case single DAC conversion from various DAC devices also cannot be detected, right?

 

[oivavoi]

To make such conclusion you should be totally sure that ADC didn't introduce any effects that can be heard. But you can't be sure in that as that has not been proven.

I would guess that they have ran repeated loops with the ADC in question, and determined that they couldn't detect it. And again, if it did color the sound, I struggle to see any logical reason why this coloration would be detectable with some devices but not with others.