Explain the non constant delay? You are right about smallest differences however it seems these smallest differences "destroy" other DACs. I suspect the R2R DAC made with .1% resistors and discrete transistors is unlikely to even meet an ENOB of 16 bits and it will be very sensitive to environmental effects. And then it sounds better? Too many contradictions. The suggestion that some other parameter that we have not identified is important may have some value but handwaving is not testing ideas. Non-stocastic noise modulation can be tested. Not easy but not impossible. Timimg modulations, dynamic amplitude intermodulations etc. can probably be tested for. Not easy.
More or less, but its not quite that simple.
Digital audio is by standard measurements almost perfect. What impressive number did I mention recently. -130dB? I think that was it.
The problem with digital audio is what I was trying to warn about: What are we measuring, and what are we not measuring?
Even today, digital audio reproduction quality continues to advance. There are remaining known problems. There problems with PCM, and there are problems with DSD.
Companies like ESS and AKM aren't fools. They aren't just selling newer and newer snake oil, when really a several year old Vega sounds just as good to every last human on earth. Who needs a Hyperstream 4 modulator when Hyperstream 2 was already audibly perfect? Why bother? Its all just a big lie to keep selling products that measure slightly better, but don't sound the slightest bit better to any human all? Is that what you think?
We are letting ourselves be fooled by measurements that don't measure everything, is what I think. We pretend that what we aren't measuring doesn't matter, that it must be negligible.
In another 10-years digital audio will probably be even closer to its claimed 'perfect sound forever.' It isn't there yet though, not IMHO anyway.
EDIT: I know this can devolve into a conversion about ABX DBT validity and various other stuff. It would be just as pointless as it always has been.
The only way forward I have been able to think of over the years I have been in this forum is for some of us to get together and listen at the same time and in the same place. I have invited people to Auburn but the few that have bothered weren't the 'digital is already audibly perfect' types.
Last edited:
Pre-pandemic, a friend came around with his latest digital gear. This guy is no fool - he is involved in high speed digital development (non audio). His audio gear was semi custom.
Anyway, my old Logitech Transporter sounded to both of us significantly better. He left rather downhearted.
Year 1 of the pandemic he was back. So both of us sporting masks we unboxed his latest digital gear. It was night and day - it now sounded so much better than my Transporter it was a joke.
We spent the rest of the time listening to different digital filters that he had incorporated into his gear, They all gave a different version of the same music files.
So agreeing with Markw4 here - there is still much to learn in perfecting reproduction of digitally stored music.
FWIW I now use a Cambridge Audio CXN V2, and am very pleased with its sonic performance. And sooo much easier to drive than the decades-old technology of the Transporter.
Anyway, my old Logitech Transporter sounded to both of us significantly better. He left rather downhearted.
Year 1 of the pandemic he was back. So both of us sporting masks we unboxed his latest digital gear. It was night and day - it now sounded so much better than my Transporter it was a joke.
We spent the rest of the time listening to different digital filters that he had incorporated into his gear, They all gave a different version of the same music files.
So agreeing with Markw4 here - there is still much to learn in perfecting reproduction of digitally stored music.
FWIW I now use a Cambridge Audio CXN V2, and am very pleased with its sonic performance. And sooo much easier to drive than the decades-old technology of the Transporter.
My advise to anybody claiming to hear differences in well-measuring dacs is to do the following:
If you still can confidently pick which is which, login to ASR and post your findings there. AFAIK they have a reward. Easy money just waiting to be picked up.
- Match levels to 0.1dB
- Eliminate sighted observation (i.e. not knowing which is playing)
If you still can confidently pick which is which, login to ASR and post your findings there. AFAIK they have a reward. Easy money just waiting to be picked up.
That’s exactly what I mean. Don’t bother about the why and how, just listen without prejudice.
There are more than just a few peope who prefer SE tube amps, with THD in the percents, over a SS amp with THD below -120 dB.
Same ratio in distortion for vinyl versus digital, still preferred by large amounts of people.
Trying to explain things leads to never ending discussions while turning in circles all the time, without making any progression just like Howie mentioned.
Hans
I especially like this:
"I won't speculate on the audibility of this phenomenon but anything that is measurable is fair game for me. If people are going to shell out serious moolah for a DAC, least thing you can do is show an objectively provable benefit."
That mirrors my thinking as well on this subject even though I'm just a hobbyist.
I totally agree and just for these reasons there should be no need to make sweeping claims about something sounding better than something else.
Something interesting about digital filters in dacs: Some people don't hear the effects of dac digital filters. IIRC some people gone so far as to claim that the idea that the filters have any audible effect is a marketing lie told by dac manufacturers. That anybody who thinks they hear a difference in the filters is imagining something that isn't real.
Changing the subject for a moment, there is EE who goes by the forum name of PMA here and at ASR. He is very much a measurement guy. He used to put on listening tests here in this forum. He would make some recordings and ask people to use Foobar2000 ABX plugin to show if they could actually hear a difference. The results were typically not very good. Nobody could pass blind ABX with any consistency.
Yet PMA said he had a friend who was a very good professional symphonic musician, and that PMA knew the guy could hear things PMA couldn't hear.
Anyway, the listening tests by PMA continued until finally PMA found he was starting to learn how to pass blind ABX. When he started to be able to do that he made a very brief post about it, using two words straight out of the so-called 'ABX denial playbook.' I remember that phrase because I was accused of speaking 'straight out' of that 'playbook' before when I used the same words that PMA did.
For a little fun, maybe try to guess what the two words are in PMA's post:
I got 5/5 from xyz vs. 123 ABX. Next run was a mistake + fatigue. Too much concentration for ABX first five 
. XYZ is "less colored".
https://www.diyaudio.com/community/threads/subjective-listening-test.273938/post-4324842
. XYZ is "less colored".https://www.diyaudio.com/community/threads/subjective-listening-test.273938/post-4324842
Last edited:
How about something else PMA said:
"...if someone has achieved any valuable result with foobar ABX, it was with large, matured speakers."
https://www.diyaudio.com/community/threads/subjective-listening-test.273938/post-4324530
EDIT: BTW don't think anyone claimed any audible difference between two dacs is exclusively caused by HD/THD. How about 'signal-correlated noise' that doesn't show up well on an FFT? How about random clock jitter that affects imaging but doesn't show up on J-Test?
"...if someone has achieved any valuable result with foobar ABX, it was with large, matured speakers."
https://www.diyaudio.com/community/threads/subjective-listening-test.273938/post-4324530
EDIT: BTW don't think anyone claimed any audible difference between two dacs is exclusively caused by HD/THD. How about 'signal-correlated noise' that doesn't show up well on an FFT? How about random clock jitter that affects imaging but doesn't show up on J-Test?
Nobody has claimed that HD/THD or SINAD (or anything else you may come up) is the cause. But the first step to advance science is to ascertain that the difference is really audible. This is why level matching and blind testing is essential.
To repeat:
http://seanolive.blogspot.com/2009/04/dishonesty-of-sighted-audio-product.html
In case you missed it I already explained that the use of the term 'destroy' was intended as the taking of some literary license for the purpose of emphasis. It was not intended to be taken as literal. I will tell though, for me personally there are some dacs that if they are all have then I won't listen to music at all because I can't stand the sound, even for background music. A number of them have come through here. I would rather listen to dac in my laptop in some cases, because at least it hides most of the intolerable 'garbage' under smooth resistor noise.
Not to me, not so far anyway. A discrete resistor DSD FIRDAC @DSD256 with shunt regulated Vref, etc. Yes, I can think of one that sounds better.
Agreed that some things are measurable even if hard to do. However, measurement may not be hardest thing. Showing evidence of audibility and or effects on preference in statistically significant populations can more costly in time and money. Even then, one such study could be dismissed as an outlier. Dismissal and or denial have happened many times in medicine.
Moreover even some large-scale studies have proven wrong. In many cases that's because of the huge variability in humans. A few hundred or several thousand humans may not be enough to get it right, depending.
We haven't even started to talk about experimenter bias, either.
The above sorts of problems can take multiple human lifetimes to sort out. I don't even have a whole lot of one lifetime left.
As a more practical matter what I would like to get a better handle on, using shared listening sessions on a good system, would be for a rough estimate to what extent the variations among forum members reports of audibility is related to individual differences as opposed to system differences. I already have some limited evidence that both are most likely involved and probably intertwined. Also limited evidence that listening skill is to some extent learnable in various domains (e.g. musical performance discrimination, audio system SQ discrimination, etc.).
Regarding discrimination versus preference, if you tell some one I am going to play a musical passage for you. I can repeat if for you as many times as you need to decide. There is a cymbal hit at about 5-seconds into the passage. Would you describe the cymbal sound as more: (1) rough, or (2) smooth? That is what I would consider an elementary example of discrimination as opposed to preference. I find that level matching for some discrimination tasks tends to be less critical than for preference testing, not that there is any reason not to level match.
Last edited:
Level matching is important in both discrimination and preference testing. When comparing DACs I do the matching with an ADC using various signals. Matching to 0.1dB would be quite difficult with DVM. Another thing is that most DAC or player volume controls are too coarse. I use analog potentiometers. To make the testing blind you need an ABx gadget that has truly random switching and hides all visual or audible cues regarding the switching.
Had some experience with a switched resistor volume control. Found at some point the volume control sounded somehow wrong. Turned out the relay contacts became noisy with continued use, likely wear of thin surface plating. The relays introduced what would probably be better termed as signal-correlated noise as opposed to termed as HD. Such things can be confounding factors in ABx gadgets, not that such issues can't be found and corrected.
Also found that potentiometers can create audible distortion. This is well known as due to resistive element nonlinearity and wiper contact nonlinearity. After trying various solutions found Goldpoint attenuators the most transparent option. Of course they are stepped.
Point I am trying to make is that for testing of very high performance systems, test gear can potentially cause hidden confounding.
Also found that potentiometers can create audible distortion. This is well known as due to resistive element nonlinearity and wiper contact nonlinearity. After trying various solutions found Goldpoint attenuators the most transparent option. Of course they are stepped.
Point I am trying to make is that for testing of very high performance systems, test gear can potentially cause hidden confounding.
Last edited:
Exactly - lots of things sound different, it's subjective. And good - life would be dull if we all likes the same. There's some music where I can clearly hear differences, as I like it and know it well, and other music where I can't tell the difference, all I can tell is it's horrible to me! as long as everyone remembers, no subjective view is "right".
They make new products for many reasons - some of which, like everyone else, is to excite their market and gain share. Some of it is also because they can. Some of it is because they have to - fabs no longer run the process needed, or have ramped the cost as it's old. Some of it is because they think it's better. Many other reasons. The result is all good, for us.