GoldenSounds passes apparently ABX test for DACs (NOT Really)

[amirm]

Most professional audiometry gear doesn't go that high for a reason, you have to use something special.

Indeed. I was very excited to get a full frequency response test the time before last, only to be told the limit was 8 kHz. They told me hospitals have testing up to 20 KHz and that it is used to diagnose some kinds of cancer.

 

[amirm]

First time I have heard of this. Could you explain why, please?

With ABX test, you only have to detect the identity of the sample, not its fidelity, to pass it. Let's say the two samples are identical sounding but one starts a hair later. You could use that delay to detect which is which and pass the test without any fidelity difference.

 

[Mart68]

He has had that ABX box for a while. The fact that we haven't seen him pass any such blind tests means that he likely can't pass them. What we see here may be the best he can do.

At this stage I think that's got to be the case. He's had that box a few years now.

And in fairness in this latest video he does make it very clear that sighted impressions can't be trusted. In fact nothing he says in this latest one is wrong.

So it must be that he's tried to identify the DAC differences he used to claim were there, failed, come to realise the truth.

He's clearly very bright, he presents very well, he could be an enormous asset to the hobby. I don't know if he makes a living elsewhere or if this is all he does. Shame if it is the latter as if you've got a financial stake then it's impossible to be impartial.

 

[Keith_W]

With ABX test, you only have to detect the identity of the sample, not its fidelity, to pass it. Let's say the two samples are identical sounding but one starts a hair later. You could use that delay to detect which is which and pass the test without any fidelity difference.

Sorry, I should have been more specific in my quoting. I was asking why audiometry does not test hearing above 8kHz.

 

[voodooless]

Sorry, I should have been more specific in my quoting. I was asking why audiometry does not test hearing above 8kHz.

Probably because of the same reason that analog telephones had about 8 kHz of bandwidth: it's enough to understand speech.
Audiometry is mostly about making sure people can understand each other, they are not there to fix your wildly inadequate hearing for high-fidelity audio That's what your imagination is for

 

[KiyPhi]

First time I have heard of this. Could you explain why, please?

Interesting that you mention 8kHz because that was my limit on my hearing test yesterday. In three weeks time there will be a more thorough test which will introduce me to v expensive hearing aids.

It is because at those frequencies, the individual's ears matter a lot so you need specialized equipment. Most places that test hearing do so for occupational testing or for basic hearing aids. From their viewpoints, getting that specialized equipment is mostly unnecessary. In the case of occupational medicine (where I used to work), it is because companies like the FRA are only concerned with the ear health of the "functional rage" where anything above that is considered functionally unimportant as you can get all the information you need to function regularly below 8kHz. Hearing aids also for the longest time only went up to 8kHz for the same reason, the frequencies above that weren't considered important enough to invest the money in the fancier equipment. Even now, the specialty hearing aids only go up to about 10kHz or so.

The technology to test higher has existed for a long time, headphones like the DD450 have been around since at least 2017 but require special equipment with high-cost special licenses. Most places don't want to pay those extra fees without a good justification.

 

[Keith_W]

It is because at those frequencies, the individual's ears matter a lot so you need specialized equipment.

That makes sense. I have read articles about the ridges and shape of the pinna bouncing short wavelengths and reflecting some of them away. The reason our ears are shaped thus is to deliberately modify the HF response so that we hear something different when we turn our heads - very important for determining the direction of sound. I guess that if hearing aids were to deliver a constant unchanged "correction" above 8kHz, it does not reflect how we were evolved to hear.

 

[OldHvyMec]

Even I could possibly detect those ?

You probably could if you had a way to amplify the sound which seems to be what many leave out of a test. We control the test, BUT
not all people are trained for the same type of hearing needs used in their particular jobs. At 150 DB for example you might be surprised
how easy it is to hear 18khz. or using a tool that doctors uses like a simple stethoscope. There are differences or animals wouldn't be able to
pick up mice beneath the snow or an owl hear a rabbit munching grass 75 feet away. If we reproduce their hearing mechanism or enhance our
own ears what would anyone expect? I know, I used a parabolic with great success on job sites a few time to hear which one was making the
wrong noise out of the 25 pieces running at 110db (at least.) When you're the size of an ant in comparison to what's around you, you proceed
with extreme caution. An orange vest just means you'll be a flat orange vest if your not careful.

 

[KiyPhi]

That makes sense. I have read articles about the ridges and shape of the pinna bouncing short wavelengths and reflecting some of them away. The reason our ears are shaped thus is to deliberately modify the HF response so that we hear something different when we turn our heads - very important for determining the direction of sound. I guess that if hearing aids were to deliver a constant unchanged "correction" above 8kHz, it does not reflect how we were evolved to hear.

The shape is only part of it though. We can compensate for that, it just costs more money and most places don't want to spend that money without good reason. There is some studies that show high frequency audiometry can detect hearing loss early and be used to help come up with treatments for tinnitus since that is usually high frequency related. Your average place won't be doing those things so it isn't worth the investment.

 

[MacClintock]

It is because at those frequencies, the individual's ears matter a lot so you need specialized equipment. Most places that test hearing do so for occupational testing or for basic hearing aids. From their viewpoints, getting that specialized equipment is mostly unnecessary. In the case of occupational medicine (where I used to work), it is because companies like the FRA are only concerned with the ear health of the "functional rage" where anything above that is considered functionally unimportant as you can get all the information you need to function regularly below 8kHz. Hearing aids also for the longest time only went up to 8kHz for the same reason, the frequencies above that weren't considered important enough to invest the money in the fancier equipment. Even now, the specialty hearing aids only go up to about 10kHz or so.

The technology to test higher has existed for a long time, headphones like the DD450 have been around since at least 2017 but require special equipment with high-cost special licenses. Most places don't want to pay those extra fees without a good justification.

Interesting. Just as a joking side remark, this DD450 headphone, of which I heard for the first time, is not Harman tuned (https://www.radioear.us/-/media/radio-ear/main/datasheets/datasheet_dd450.pdf):

 

[goat76]

I’m not that surprised if he could detect some differences. A year ago I did a blind test with and without upsampling and could hear the difference, the difference I could hear was that the upsampled file sounded airier with a little bit more easily heard reverb tails. The difference was minor but was fairly easy to hear when seamlessly switching between the sound files.

 

[KiyPhi]

Interesting. Just as a joking side remark, this DD450 headphone, of which I heard for the first time, is not Harman tuned (https://www.radioear.us/-/media/radio-ear/main/datasheets/datasheet_dd450.pdf):

View attachment 367153

No, testing headphones are so expensive because they, along with the tone generator used with them, have to be able to reliably reproduce the required frequencies without distortion at the required levels. The higher the frequency, the harder that is. The headphones are measured and calibrated regularly along with the machine to ensure the settings you put in on the testing panel (or what it outputs automatically with some machines) are accurate. This is especially important for testing ranges of tinnitus as one of the therapies, if I recall, is a tone masking therapy. It is also important to make sure each time you play a 12kHz tone, it is always 12kHz so results are actually comparable.

 

[solderdude]

Interesting. Just as a joking side remark, this DD450 headphone, of which I heard for the first time, is not Harman tuned (https://www.radioear.us/-/media/radio-ear/main/datasheets/datasheet_dd450.pdf):

View attachment 367153

I measured one as well. Expensive too € 950.-

It is intended for audiometry in a not extremely quiet surrounding so one can evaluate hearing without the need for an expensive silent booth.
It clamps like nobodies business (10N) and has uncomfortable plastic pads (easy to wipe clean)
The depth is just 12mm so most pinnae will be bent 'flatter' to the head.
The curve meets the RETSPL standard used in audiometry which compensates for the frequency response so in order to use it with the specialized equipment used in audiometry (with a specific compensation)

It designed to meet IEC-60645-1 and ANSI S3.6 standard for measurements between 125Hz and 16kHz

Not suited for hifi ...

 

[Hayabusa]

Probably because of the same reason that analog telephones had about 8 kHz of bandwidth: it's enough to understand speech.
Audiometry is mostly about making sure people can understand each other, they are not there to fix your wildly inadequate hearing for high-fidelity audio That's what your imagination is for

4kHz?

 

[Hayabusa]

With ABX test, you only have to detect the identity of the sample, not its fidelity, to pass it. Let's say the two samples are identical sounding but one starts a hair later. You could use that delay to detect which is which and pass the test without any fidelity difference.

So ABX test are not saying anything also?
Can we just have some positive sound for this guy doing ABX tests as we are all on this forum saying the audiophiles should do to prove they hear a difference?

 

[CedarX]

So ABX test are not saying anything also?
Can we just have some positive sound for this guy doing ABX tests as we are all on this forum saying the audiophiles should do to prove they hear a difference?

There might be some ‘technical’ issues/concerns with the details of the ABX tests he did, but to be positive, let say it’s a good ABX test…

The part I’m questioning is what he uses the ABX test for: he demonstrates that, granted his young age, exceptional hearing, a treble boost applied to an already bright HP… he can hear the difference between two filters mimicking two different DAC reconstruction filters. That’s not done with two different DACs but with two different files, same DAC/Amp/HP.
He then concludes that, since every reconstruction filter is (slightly) different from DAC to DAC, there are always audible differences between DACs.

That’s a long intro… but it’s also because he has a convoluted way to “demonstrate” his point! Why didn’t he choose two random, but transparent, DACs and perform an ABX test between the two using his standard test playlist? He has an ABX relay-based switch board and I’m sure he’s competent enough to do just that type of test (he showed the board in the video). Why did he choose this convoluted demonstration path instead? Is it because he would fail the two DAC test, leading to a different conclusion?

That’s what puzzles me… Not the ABX test itself, but the underlying purpose and use of an ABX test for what appears to be a GoldenSounds agenda.

 

[SIY]

What we see here may be the best he can do.

Maybe. If I were dishonestly trying to make a point, I could easily show a pass for that test despite having a personal HF cutoff slightly below 14kHz. It's trivial.

 

[MacClintock]

There might be some ‘technical’ issues/concerns with the details of the ABX tests he did, but to be positive, let say it’s a good ABX test…

The part I’m questioning is what he uses the ABX test for: he demonstrates that, granted his young age, exceptional hearing, a treble boost applied to an already bright HP… he can hear the difference between two filters mimicking two different DAC reconstruction filters. That’s not done with two different DACs but with two different files, same DAC/Amp/HP.
He then concludes that, since every reconstruction filter is (slightly) different from DAC to DAC, there are always audible differences between DACs.

That’s a long intro… but it’s also because he has a convoluted way to “demonstrate” his point! Why didn’t he choose two random, but transparent, DACs and perform an ABX test between the two using his standard test playlist? He has an ABX relay-based switch board and I’m sure he’s competent enough to do just that type of type of test (he showed the board in the video). Why did he choose this convoluted demonstration path instead? Is it because he would fail the two DAC test, leading to a different conclusion?

That’s what puzzles me… Not the ABX test itself, but the underlying purpose and use of an ABX test for what appears to be a GoldenSounds agenda.

GoldenSounds passes apparently ABX test for DACs (NOT Really)

I am more interested in his ABX device than DACs. Do we have info about it? haha me too - I already asked him in the comment section some hours ago

audiosciencereview.com

Maybe he was using the files generated with the different filters via software to be able to share them and prove his "point" stronger that way by making it potentially reproduceable by others. But the crucial criticism remains either way, if he passes the ABX test between the files or between two real DACs, the differences are so miniscule and have absolutely nothing to do with the ones often purported by "DAC audiophiles", like more air, better imaging. larger soundstage, better resolution you name it. Just some more content at around 20kHz detectable only with supposedly ear-hurting sound levels and special bright headphones that only a tiny percentage of the population is able to hear at young age. Well, that is clearly a justification to throw out a few thousand bucks for a "better" DAC!

 

[Arnas]

Yes, he does have a AP5xx, and, being the person that he is, made a video about how measurements can be misleading......

I hate to call someone names, but he is moron for buying TOTL measuring gear while being sujectivist.

 

[solderdude]

Some replies from Cameron (GoldenSounds) in answer to some questions in his comments.

The two files are to test the effect of the oversampling filter in the DAC. By showing that this single factor makes a difference, that shows that DACs make a difference (and oversampling as a standalone factor makes a difference). The reason to do it this way is because there is no way to do a physical ABX test in a remotely verifiable way. Doing a digital test with shared and verifiable inputs and results was a necessary precursor to sharing physical ABX results in future reviews as this establishes a conclusive difference in a way that does not rely on trust.

I have various expensive power cords, I have never heard them make any difference, nor is there any explanation as to how they could unfortunately. Power conditioning is one thing, power cables are another

I've also never heard cable differences

Whether there is a difference for most people, or enough of a difference for the average consumer to warrant spending money on is a separate question. This was intended to answer the question "Can well measuring DACs sound different", to which the answer is conclusively yes

'm not saying it is. I and others use sighted testing all the time, and sighted testing is often fine when you're comparing things that are immensely different with clear and credible explanations and evidence as to why. But if you're trying to answer whether something does/does not sound different, you cannot conclusively answer that by sighted testing

This video does not and was not intended to discuss all the numerous other factors. But simply to answer the question "Can well measuring DACs sound different". Interpretation or evaluation of all the other factors is outside the scope of this video.

About placebo:

It can be shockingly powerful. One time I was comparing two EQ profiles, had my mouse over the button toggling on/off and trying to decide which I liked more. After a while I looked up....aaand the mouse wasn't even over the button, nothing was changing. It can be quite humbling when you have an experience like that and realise how massively your perception can be influenced by your expectations.

It can yes, but not in a consistent manner. Your brain can trick you into thinking one file sounds a certain way, but the point is that if you do enough runs, you can then show whether that difference is consistently attributable to one of the devices/files (ie: there is a genuine audible difference you can reliably pick out), or whether it was random and not likely a genuine difference

Interesting conversation:

@mccririck01

Why not compare a topping DAC to your May instead? Both ran with the same upsampling filter.

@GoldenSound I have done, if they were the same I'd have sold the May and freed up a bunch of cash

@mccririck01

Yeah that's a good point. @GoldenSound why not video the abx then?

@GoldenSound
@mccririck01 I'll happily post some, but as discussed in this video, the problem is that people will just say "Ehh you faked it/you're lying". That's the entire reason why this test was done in a remotely verifiable way

@sebastiantomita5956
@GoldenSound I meant to compare them in the same manner and make it public. You said you couldn’t find a valid published test. Well, you haven’t changed that, have you? You just tested the effect of oversampling. That’s less important than the DACs themselves. Just my 2 cents.

@GoldenSound
@sebastiantomita5956 I wanted to have this video out first

Looks like he did ABX different DACs too but for some reasons did not post them (yet ?)