can DACs sound different if they both measure well?

[acko]

DACs regularly clip on music, which they should not do in my opinion.

To check whether it is audible on music requires an extensive double-blind test with carefully selected music and carefully selected listeners, which test no one has done as far as I know. It would not surprise me if it were audible on some recordings, but I simply don't know as no one has done that test.

I think this concern is more with other type of DACs. DACs built around this chip with digital vol adj happening before filters at normal listening levels not affected and no further tests needed

 

[MarcelvdG]

How do these supposed psychological effects impact e.g an ABx test? Are you (or Markw4) suggesting that people who don't believe in audible differences somehow fool themselves into getting worse results?

Precisely as Mark just wrote:

For people who expect to hear no difference, maybe they will hear what they expect even if there is some slight audible difference they tend not to notice. For gross audible differences that should more likely come through despite expectation to the contrary.

You need people who either think they hear a difference or at least have an open mind about it.

 

[Miller-8]

You need people who either think they hear a difference or at least have an open mind about it.

Not entirely true. You could add a component that definitely changes the sound in a measurable way (frequency response change) that they have to look out for as well as trying to tell the difference between two unaltered DACs.

Also you could have half that believe they can tell a difference and half that dont.

 

[MarcelvdG]

I think this concern is more with other type of DACs. DACs built around this chip with digital vol adj happening before filters at normal listening levels not affected and no further tests needed

This one should be fine if you actually use the digital volume control.

 

[bohrok2610]

You need people who either think they hear a difference or at least have an open mind about it.

Most of us have an open mind about it so that argument fails. Why would anybody not having an open mind about audible differences run an ABx test?

 

[bohrok2610]

DACs built around this chip with digital vol adj happening before filters at normal listening levels not affected and no further tests needed

All AKM AK449x DACs (with the exception of AK4495) have attenuation control before filter.

 

[acko]

This one should be fine if you actually use the digital volume control.

Noted, thanks. Very interesting to know actually 🙏

 

[acko]

All AKM AK449x DACs (with the exception of AK4495) have attenuation control before filter.

Yes, they are also and end users will not notice anything unusual if they use digital vol only or do some similar adjustments in ‘preamp’ mode

 

[MarcelvdG]

By the way, in the echo test, out of the six participants, the one who wasn't sure he heard any difference at all (but had an open mind about it) was the only one with a significant result.

 

[Markw4]

I have tried ABX as PMA requested. I found I could do better by sorting his files in order of distortion blind, ears only. It was hard, but easier than ABX.

I have also tried A/B, which I also find to be easier than ABX.

Thus, I have no idea what the fixation about ABX is all about. Why use a protocol that tends towards false negatives more than other protocols? Is it because the experimenter wants to see people fail due to poor experimental design?

Even considering the above, I made a few recommendations on how to make Foobar2000 ABX more fair to test subjects. Its only takes a few simple changes. There was no way to contact the guy who wrote it; much like the mysterious bitcoin guy is what it feels like.

 

[MarcelvdG]

Most of us have an open mind about it so that argument fails. Why would anybody not having an open mind about audible differences run an ABx test?

For example, I've done a slow-versus-fast-common-mode-loop test with an ABX switching box on the shift register FIRDAC because Mark asked me to listen to the difference between a slow and a fast loop. I was quite sure I would not hear any difference. I indeed didn't hear any difference, but that result means absolutely nothing.

 

[MarcelvdG]

I have also tried A/B, which I also find to be easier than ABX.

How is a blind A/B test done?

 

[Markw4]

IMHO its similar to ABX, but instead of "X is A and B is Y or else Y is A and B is X," there is a button to play A, a button to play B, and a button to play Unknown (which will be either A or B for this trial)). Listen to them until you are ready to vote for the Unknown sounding more like A or more like B.

There is also the issue of forced choice to consider. If you have done a few trials and you are getting fatigued, IMHO you should be able to have a button that says "Decline to Vote" for this trial. I don't want to start getting fatigued then be forced to guess. I don't want to guess unless I choose to do so willingly of my own volition. Personally, I am not a gambler. I would rather sit out a round if I feel like my acuity is starting to fade.

So, if I decline to vote and say I will come back tomorrow to do some more trials, but when I get here tomorrow I want to start with a few practice trials before jumping back into the test. Why shouldn't I be able to do that? The only reason I wouldn't be allowed to would be because it is inconvenient for the experimenters. Well, seems to me that's too bad for them. Do they want to know if I can reliably hear a difference or not under normal circumstances where I am not fatigued? Or would they rather force me to fail from listening fatigue because its more convenient for them?

 

[mdsimon2]

there is a button to play A, a button to play B, and a button to play Unknown (which will be either A or B for this trial)). Listen to them until you are ready to vote for the Unknown sounding more like A or more like B.

There is also the issue of forced choice to consider. If you have done a few trials and you are getting fatigued, IMHO you should be able to have a button that says "Decline to Vote" for this trial. I don't want to start getting fatigued then be forced to guess. I don't want to guess unless I choose to do so willingly of my own volition. Personally, I am not a gambler. I would rather sit out a round if I feel like my acuity is starting to fade.

So, if I decline to vote and say I will come back tomorrow to do some more trials, but when I get here tomorrow I want to start with a few practice trials before jumping back into the test. Why shouldn't I be able to do that? The only reason I wouldn't be allowed to would be because it is inconvenient for the experimenters. Well, seems to me that's too bad for them. Do they want to know if I can reliably hear a difference or not under normal circumstances where I am not fatigued? Or would they rather force me to fail from listening fatigue because its more convenient for them?

This is exactly how the Deltawave comparator works. Button to play A, button to play B and button to play X. Switch between them as many times as you want until you are ready to say if X is A or B. No time restrictions.

Does this mean we can expect you to post your Deltawave comparator results for @bohrok2610's DAC recording and the original?

Michael

 

[Markw4]

Does this mean we can expect you to post your Deltawave comparator results for @bohrok2610's DAC recording and the original?

Do you or bohrok2610 claim that the original and one of the DAC->ADC copies are audibly transparent from each other? bohrok2610, do you so claim?

Do they look the same in deltawave?

In case its not clear, I am starting to get a little interested in the subject.

 

[MarcelvdG]

there is a button to play A, a button to play B, and a button to play Unknown (which will be either A or B for this trial)). Listen to them until you are ready to vote for the Unknown sounding more like A or more like B.

I thought that was what was meant by ABX. Apparently I had the terminology wrong. In post #891, ABX should be AB then.

 

[Logon]

So, we have one cube and one ball, made of the same substance, that measure equal by weight and volume, so they are indistinguishable on the ultimate measuring devices for weight and volume.

This is precisely why we need humans to evaluate things, everything.
And listen to it, if it is an audio device.
The measurement, as you imply, are useless without humans to validate their meaning in the context for which they were created.
At the moment we have no measurements that tell us how a device sounds, other than our own ears.

 

[jean-paul]

All AKM AK449x DACs (with the exception of AK4495) have attenuation control before filter.

So a non issue in such cases?

Open question: why are rare or even non occurring "problems" so highlighted? If these do not occur with recent ICs it is not worth marking these as issues.

 

[Logon]

It was like that with “That signal tells everything about the "sound" of the device „ (you didn't note that bohrok2610 has put quotes around word sound, and that this translates as „ DAC output signal contains absolutely all information that can be related to sound”.

This seems to me still philosophy.
It "contains absolutely all information" put that way it seems like we know everything about what's inside the electrical signal.
And it sounds like you really believe that.
I don't, I don't believe we know everything about that electrical signal.
I think everything is there, but we don't know what exactly that everything is.

In other words, if everything is in the electrical signal, but we don't know how to extract the part that tells us how a device sounds or even what kind of thing that part is, what are you two talking about?
Perhaps you two meant to say that everything is in the electrical signal is there, but we only know a small part of it, so small that it tells us a little more than distortion, noise and frequency response?

And when I said that it should not be said that of a DAC we measure the sound, but the electrical part of the sound, didn't I perhaps say the same thing as above?

All the measurements taken relate to the sound and distortion of the sound. And it's the sound that matters

This doesn't seem to me to be said correctly.
Regardless of the fact that I don't even know what SINAD is because I'm not interested at the moment, it seems clear that the measurements from wherever they come from are related to the devices (that's to the "electrical" part of the sound), not to the sound.

Now there's a misunderstanding. DACs do not output sound but only an electrical signal. That signal tells everything about the "sound" of the device

Statements above have exactly the same meaning.
I don't see any misunderstanding.

No they don't. As I said the electrical signal tells you everything about the sound.

Really?
Where did you read that?

Please enlighten us how a DAC can impact the sound through other means than the electrical signal it outputs.

It seems clear that the following useless (and certainly not the only one of that kind) "conversation" above is based on something wrong.
And it is not something related to the electrical signal coming out of a device.

Saying that there is "everything in electical signal about the sound" furthermore with such emphasis implies that that everything is also related to how a device sounds, but obviously that is not the case, and this is also very misleading.

The measurement that tells us how a device sounds has not yet been discovered, even if philosophically speaking there is already everything in the electrical signal.

 

[bohrok2610]

Do you or bohrok2610 claim that the original and one of the DAC->ADC copies are audibly transparent from each other? bohrok2610, do you so claim?

That test was initiated by you claiming to hear a clear difference between 2 versions of PCM2DSD and preferring the older version that in measurements was worse. So the comparison was supposed to be between recordings made with those 2 versions, not between original and any one of the recordings. What is worth mentioning is that later you changed your mind and agreed that the later version is better. That tells a lot about the reliability of sighted listening tests.

But of course you are free to try ABx comparisons between the original and one of the recordings. Based on your claims you should have no trouble with that one.