GoldenSounds passes apparently ABX test for DACs (NOT Really)

[Thomas_A]

Done testing. I did not find any tells this time either. I would really like to see this 20-trial experiment repeated in 10 successive sessions. If anyone of them get 18/20, I might believe it.

 

[MacClintock]

Again as I noted very early in the thread, controlled blind testing by Meridian/MQA folks showed audibility of different DAC filters. This was peer reviewed and published at AES.

I read your summary and the discussion about it, unfortunatley AES is such an exclusive club that the article itself is nowhere in the internet available. But the question treated in this article is a bit different, it is about downsampling a high-res 192kHz, 24 bit file to 16 or 24 bit and 48 or 44.1 kHz. I am not sure if these two scenarios are so easily comparable, as in our case here, there was no high-res content initially.

Furthermore Bob Stuart made several dubious claims, like "that the sounds that are important to us are not represented by the frameworks described by Nyquist and Shannon. Firstly, he emphasises that sound is not inherently band-limited and hence there are existent audio signals that extend beyond the documented human threshold of hearing.", see https://www.idrumtune.com/high-resolution-audio-how-true-is-your-playback/
Adding the MQA fame, I would take everything from him with a grain of salt, to say the least, as he clearly seems to have an agenda.

 

[Geert]

I am not sure if these two scenarios are so easily comparable, as in our case here, there was no high-res content initially.

There's indeed no content in what we typically consider high res, but there is in the 20kHz-22kHz region where the response of the 2 filters differs. Pkane's analysis of >20kHz: "Peak difference of -74dBFS, RMS of -103dB. Much, much smaller, but still appears somewhat correlated to original soundtrack"). So to understand if the AES test is comparible, we would need to see a spectral analysis of the track they used for their test (Haydn's String Quartet Op.76 No.5 in D “Finale, Presto" from “Nordic Sound (2L Sampler)"). Any volunteers?

An interesting observation from the AES test (from Amir's write up) :
"Let's note that none of the testers could hear above 20 KHz and some probably could not even get that far. How is it that filtering the ultrasonics that testers could not hear was audible? The answer is that we introduced audible artifacts, not that we removed what we could not hear in the first place.The paper authors hypothesize that it is the filtering “ringing” in the time domain that may have caused these artifacts. Ringing is an unavoidable manifestation of digital filtering. The sharper the filter has to be, the longer its “tails” (ringing) in time domain. The ear is not a spectrum analyzer but rather "hears" the samples as they arrive. In theory the ringing can be audible and potentially is what was heard".

 

[Tell]

I think we should focus the discussion on the pertinent facts and data, not persons. I, personally, dislike GoldenSound quite a bit, but he apparently did something nobody expected to be possible and it is not completely clear how he did it, so it is interesting to discuss it and elucidate the scientific and technical basis of it.

Yes of course pertinent facts and data is what matters, I also have some issues with GoldenSound but I do think that he didn't do any obvious cheating here seeing that he had so many controls for us as an audience, while Sharur on the other hand recorder a small part of his screen and nothing else, so given his track record of bullshitting in the past I wouldn't doubt for a second if he had a live spectrum analyser or whatever on his other screen, so I wouldn't call his test pertinent in any way.

 

[voodooless]

Wasn't that AES paper highly criticized?

If ringing in the hypothesis, why not prove it? Finding ringing must be trivial, right? It should show up in a simple spectrum analysis of the diff, or even looking at the the waveforms compared to the original. Some food for thought:

Audiophile Myth #260: The Detestable Digital Filter Ringing and Real Music...

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

Let me answer my own question: because the answer is in the way of pushing nonsense technology like MQA...

 

[OldHvyMec]

I think the title should be changed. (NOT Really) needs, to go. Just sayin'. Does it impact the music or the ability to enjoy it in any way other than
to argue if some person can or can't hear a difference.

My conclusion is if you turn up anything loud enough, you'll hear it all right. Right up to the point of being turned from a semi-solid into a vapor and a pile of bones.

Regards

 

[amirm]

(NOT Really) needs, to go

??? No DACs were tested as originally claimed.

 

[JSmith]

(NOT Really) needs, to go.

"Not really" was added to the title, as 2 DAC's were not actually compared.


JSmith

 

[Sokel]

There's indeed no content in what we typically consider high res, but there is in the 20kHz-22kHz region where the response of the 2 filters differs. Pkane's analysis of >20kHz: "Peak difference of -74dBFS, RMS of -103dB. Much, much smaller, but still appears somewhat correlated to original soundtrack"). So to understand if the AES test is comparible, we would need to see a spectral analysis of the track they used for their test (Haydn's String Quartet Op.76 No.5 in D “Finale, Presto" from “Nordic Sound (2L Sampler)"). Any volunteers?

An interesting observation from the AES test (from Amir's write up) :
"Let's note that none of the testers could hear above 20 KHz and some probably could not even get that far. How is it that filtering the ultrasonics that testers could not hear was audible? The answer is that we introduced audible artifacts, not that we removed what we could not hear in the first place.The paper authors hypothesize that it is the filtering “ringing” in the time domain that may have caused these artifacts. Ringing is an unavoidable manifestation of digital filtering. The sharper the filter has to be, the longer its “tails” (ringing) in time domain. The ear is not a spectrum analyzer but rather "hears" the samples as they arrive. In theory the ringing can be audible and potentially is what was heard".

Not the specific one but typical sample of their stuff.

TRONDHEIM SOLISTENE - Divertimenti - Pure Audio Recordings

Divertimenti were composed for various social occasions and were intended to be light, uncomplicated and cheerful.

pureaudiorecordings.com

(not the version of the page,the one analyzed is the 24/352.8Khz version) :

 

[Geert]

So to understand if the AES test is comparible, we would need to see a spectral analysis of the track they used for their test (Haydn's String Quartet Op.76 No.5 in D “Finale, Presto" from “Nordic Sound (2L Sampler)"). Any volunteers?

Quick and dirty spectral analysis of 2 seconds (high pitched violins) from the 24-Bit/44.1 kHz version of this recording:

Lots of energy >20kHz. For this particular piece peaking at about -54dB compared to the mid range frequencies.


(2 sec export of this piece in attachment).

 

[fin]

I just want to get an idea about the differences so I use Fabfilter Q3 to apply a low cut (brick wall) at 19kHz to both files, subtract the waves to get their difference in Audacity, slow down to 0.5x speed and boost like 40db so it becomes audible. Not sure if I'm doing it right, It's just noise or somehow remotely related with the pop sound in the original music.

The original diff files without applying high pass filter however, will reveal some original music infomation. Seems there are no serious differences in group delay or phase shifts here in the lower frequency part based on the DeltaWave output so I was wondering where it came from.

 

[MacClintock]

Wasn't that AES paper highly criticized?

If ringing in the hypothesis, why not prove it? Finding ringing must be trivial, right? It should show up in a simple spectrum analysis of the diff, or even looking at the the waveforms compared to the original. Some food for thought:

Audiophile Myth #260: The Detestable Digital Filter Ringing and Real Music...

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

Let me answer my own question: because the answer is in the way of pushing nonsense technology like MQ

Nice article. So the pre-ringing and pre-echo "issue" appears largely to be exaggerated and marketing. The intermediate filter he proposes, as a mixture of linear and minimum phase, is similar to what is implemented in most DACs nowadays, as far as I can see. However, he makes one remark that is kind of a premonition:

"Stick with a high quality digital filter that is: linear phase, relatively steep that doesn't roll-off in the audible spectrum, and provides overhead protection against clipping/overload. Something like Rob Watts/Chord's accurate high-tap sinc filter ... will be best for [high-quality] recordings from the perspective of highest fidelity. As I said in that previous post, I don't think a very steep filter is necessary, something like 95% bandpass with SoX is more than good unless you are sure your hearing acuity extends beyond 21kHz ."

 

[MaxwellsEq]

Ideally the filter should result in everything over 22.05kHz being suppressed to better than -100dB

 

[Ken Tajalli]

Not the specific one but typical sample of their stuff.

TRONDHEIM SOLISTENE - Divertimenti - Pure Audio Recordings

Divertimenti were composed for various social occasions and were intended to be light, uncomplicated and cheerful.

pureaudiorecordings.com

(not the version of the page,the one analyzed is the 24/352.8Khz version) :


View attachment 368145

Help me out here, where am I going wrong:
- from 0-5 secs there is just a hash (noise) peaking about -60dB.
- there is a ton of HF noise past 50kHz, since this is DSD, had you turned off the filtering?
- Any information over 20kHz is at mostly -70dB or so, could that be just correlated noise?

 

[VientoB]

I don't know the channel. Couldn't he just be faking all this to get the community's attention to his channel (engement?)
As has been said before, it is actually not hard to fake these results with a real-time analysis in the chain

It’s not *his* channel. He doesn’t run The Headphone Show. He just does some work for them.

 

[VientoB]

Those differences are not above 20kHz. In this case they are and on top of that very short lived. Music is something entirely different than test tones and the 20kHz content is part of music that has content below 20kHz as well. To actually perceive a 20kHz you really need to crank the volume up (with test tones > 20kHz) to at least 80dB SPL with young ears.

That’s a good point. Does GS do all of his subjective comparisons above 80db?

 

[Sokel]

Help me out here, where am I going wrong:
- from 0-5 secs there is just a hash (noise) peaking about -60dB.
- there is a ton of HF noise past 50kHz, since this is DSD, had you turned off the filtering?
- Any information over 20kHz is at mostly -70dB or so, could that be just correlated noise?

That's not DSD,is the original DXD (PCM 24/352.8Khz) recording,you can see it down the info in the page I posted.
Here's how it looks in Spectrum view where you can see the noise shaping and everything:

 

[virtua]

I'd rather preserve my hearing than crank my music loud enough to hear the minuscule (and not important) contents above 20khz, if it's still even there and the mastering engineer hasn't already cut a significant amount above about 16 to 17khz (let alone 20khz) to get an increase in headroom. None of this particularly matters though as Golden Sound is more than likely falsely using this as part of some justification for actually thinking some DACs sound smoother and warmer than others or whatever he peddles on his pages. I largely see him as an objectivist concern troll because of this. Technically accurate, yes. Greatly important to sound to most people? not really. If anything good comes out of this, it'll be Chinese brands seeing it as a new challenge and optimising their filters, hopefully after addressing the hundred other more important things people have been asking for like EQ and DSP.

 

[solderdude]

That’s a good point. Does GS do all of his subjective comparisons above 80db?

That 80dB SPL is needed to 'hear' a 20kHz+ tone. In music, when 20kHz+ is present the peak SPL (in the used file) is 40dB louder so there would need to be 120dB SPL peaks when listening to that music.
To do that without any distortion (92dBV sensitivity) you will need 25V = 10W in 60ohm.
It is not a pleasant thing to listen that loud so the 20kHz+ content, most likely, will have been much lower than 80dB.

 

[dasdoing]

Wasn't that AES paper highly criticized?

If ringing in the hypothesis, why not prove it? Finding ringing must be trivial, right? It should show up in a simple spectrum analysis of the diff, or even looking at the the waveforms compared to the original. Some food for thought:

Audiophile Myth #260: The Detestable Digital Filter Ringing and Real Music...

A blog for audiophiles about more objective topics. Measurements of audio gear. Reasonable, realistic, no snakeoil assessment of sound, and equipment.

archimago.blogspot.com

Let me answer my own question: because the answer is in the way of pushing nonsense technology like MQA...

sample rate conversion artifacts of many algos are documented here: https://src.infinitewave.ca/


and yea, it's different since we don't have the aliasing artifacts in this test. we are simply comparing two low-pass filters afaiui