DAC audible differences are finally proven with ABX?

[Thomas_A]

Yes, the file posted by @Thomas_A has been wrong (it is always good to double-check before posting data).
View attachment 549588

I think that the author wanted to make this:
View attachment 549589

Below I am attaching the correct file:

Yes, that file I posted was another experiment (group delay, filter etc). Sorry for that, but it is audible as well.

Never mind, the polarity is no longer an issue in the DAC trials , it seems.

 

[pjug]

Since Josh is not supplying recordings, is there someone here who would record a few dacs (use sharp linear filters) with a good ADC at 96KHz so we can just how small the differences are? Or point me towards existing files like these. I can't find anything like that; the ones I can find online are 44.1K. I would do it myself but I am hoping for comparison of better DACs than I have.

 

[kaintfm]

I do genuinely care about the organization. I’m a monthly donor to the SPCA, as well as another animal rescue — and was long before any of this. So this isn’t for show. I’d be thrilled if this silly ABX controversy helped them.

Yeah, I find it hard to believe anything you write at this point. A normal response would have been, "Sure, I'll loan it to you if you pay the shipping." or "Sorry, no. I don't loan out equipment. However if you're interested I'll sell it to you for cost+shipping" and finally, "Sorry but I don't loan out equipment and don't want to sell the box."

This need for additional payment in the form of donations to a charity is just whackadoodle weird nonsense. It signals to me that you're either hopelessly socially awkward and have no idea how weird it sounds, or you're dragging this out for the attention and/or to have a laugh.

 

[pma]

Since Josh is not supplying recordings, is there someone here who would record a few dacs (use sharp linear filters) with a good ADC at 96KHz so we can just how small the differences are? Or point me towards existing files like these. I can't find anything like that; the ones I can find online are 44.1K. I would do it myself but I am hoping for comparison of better DACs than I have.

I could prepare something. But by now, all similar attempts that I have already done resulted in no positive abx result. And, we also have uncertainty of user’s system setup, like this earlier today. The tests are uncontrolled.

Frankly, to me it makes more sense to test DAC output against original file data. Even then the result is most likely negative.

 

[pjug]

I could prepare something. But by now, all similar attempts that I have already done resulted in no positive abx result. And, we also have uncertainty of user’s system setup, like this earlier today. The tests are uncontrolled.

Frankly, to me it makes more sense to test DAC output against original file data. Even then the result is most likely negative.

Thanks. If you have time I would still be interested in doing it for two or three DACs. I am more interested in looking at the differences in DeltaWave than doing ABX, although maybe I would try ABX after that.

Also, comparing to the original is difficult because of the HP filter in the ADC (I am assuming you would use your Cosmos). It would be interesting to try to remove that effect, though, either by using Paul's tool to undo the effect, or doing it outside of DW using filter based on the R and C on the Cosmos.

 

[Josh83]

Yeah, I find it hard to believe anything you write at this point. A normal response would have been, "Sure, I'll loan it to you if you pay the shipping." or "Sorry, no. I don't loan out equipment. However if you're interested I'll sell it to you for cost+shipping" and finally, "Sorry but I don't loan out equipment and don't want to sell the box."

This need for additional payment in the form of donations to a charity is just whackadoodle weird nonsense. It signals to me that you're either hopelessly socially awkward and have no idea how weird it sounds, or you're dragging this out for the attention and/or to have a laugh.

Nobody’s obligated to resell their possessions at cost just because someone asks. I made clear from the beginning that I wouldn’t loan it out but might sell it for “an offer I can’t refuse.” That was framed as my trying to profit from the situation, so I’ve responded by proposing that any profits go to charity. That’s a genuine attempt to address the objection, not a bad faith offer. If @amirm accepts, I’ll follow through.

 

[Thomas_A]

I could prepare something. But by now, all similar attempts that I have already done resulted in no positive abx result. And, we also have uncertainty of user’s system setup, like this earlier today. The tests are uncontrolled.

Frankly, to me it makes more sense to test DAC output against original file data. Even then the result is most likely negative.

As you say, and since polarity flip is no longer an explanation, the only way is to compare outputs of one or two of the DACs.

 

[Thomas_A]

Just to finish off, with easy files for polarity switch. My ears still a bit ringing from the multitude ABX...these are much friendlier and very easy. In essence, it can be a point to address for equipment - since some do flip polarity.

foo_abx 2.2.3 report
foobar2000 v2.25.10
0126-08-06 19:57:49

File A: Sawtooth pol flip.wav
SHA1: 4f1e5e65b6f5aa522f74972bb2e8a0d7c4273515
File B: Sawtooth.wav
SHA1: 2662efe99a9fc33a7d73c1f03e128a0795e8abd6

Output:
Core Audio : Default, 16-bit
Crossfading: NO

19:57:50 : Test started.
19:58:28 : Test restarted.
19:58:28 : 01/01
19:58:36 : Test restarted.
19:58:36 : 02/02
19:58:43 : Test restarted.
19:58:43 : 03/03
19:58:58 : Test restarted.
19:58:58 : 04/04
19:59:05 : Test restarted.
19:59:05 : 05/05
19:59:10 : Test restarted.
19:59:10 : 06/06
19:59:15 : Test restarted.
19:59:15 : 07/07
19:59:20 : Test restarted.
19:59:20 : 08/08
19:59:26 : Test restarted.
19:59:26 : 09/09
19:59:30 : Test restarted.
19:59:30 : 10/10
19:59:35 : Test restarted.
19:59:35 : 11/11
19:59:49 : Test restarted.
19:59:49 : 12/12
19:59:56 : Test restarted.
19:59:56 : 13/13
20:00:03 : Test restarted.
20:00:03 : 14/14
20:00:07 : Test restarted.
20:00:07 : 15/15
20:00:14 : Test restarted.
20:00:14 : 16/16
20:00:14 : Test finished.

----------
Total: 16/16
p-value: 0 (0%)

-- signature --
53e283a99f709b61155712e92a9a3e1043d9c14f

Files:

Dropbox

www.dropbox.com

Dropbox

www.dropbox.com

 

[Mort]

I’ve responded by proposing that any profits go to charity. That’s a genuine attempt to address the objection

Hey welcome to planet Earth. Demanding a payment to your favorite charity on top of getting fully reimbursed for a possibly defective device is not by any measure a 'genuine attempt.'

 

[Madlop26]

yeah, this is silly, an investigator cannot do good scientific research on themselves because this creates a deep bias, the claim has high stakes and requires same level of scientific method, publishing this is irresponsible.

 

[pma]

As you say, and since polarity flip is no longer an explanation, the only way is to compare outputs of one or two of the DACs.

And without disclosing anything before the ABX results are collected. However, I am a bit skeptical that people would be willing to participate and get rid of the sighted listening comments. To prepare the test needs efforts and without enough paricipants it is not worth the time spent.

 

[krabapple]

I was never able to get correct spectra from Audacity by increasing FFT size. The result was a loss of amplitude at higher frequency components. If you are able to set Audacity to get the spectrum results as I have shown in post #899 (above yours), I will be glad to learn something. All the analyzed files are available on links that I had posted.

That's good to know. I've certainly never compared the spectral data outputs of Audacity vs Audition, so it wouldn't have occurred to me, but that's something I can try.

 

[krabapple]

Looks like he does not want to part with it which is understandable given how much time and effort he invested in obtaining one and testing with it.
Josh can ask whatever he wants as it is a rare device and that's how the market and demand works.

Nah.... this is a no win situation regardless of the effort, time and money that will have to be spent on this... just to satisfy a group of people reluctant to find the test results suspect (in the name of 'scientific truth finding').

Or, there's a deep-pockets ASR reader who's willing to meet Josh's terms (including the optional Sandy Denny box, lol, jeez), hand the damned thing over to Amir for testing, and let it all move forward?

(Mine aren't that deep, alas)

 

[krabapple]

One more time... He hasn't done a thing I asked for. What I asked for was very little: loan the box for me to measure. I also offered $10,000 insurance if it got lost or damaged.

He refused.

That's the bad look. Please don't dare putting this at my feet. The man has no ethics to demand nearly $4,000 for a box he paid $1,200 for, especially when we look at the need for the community to be served by this kind of analysis.

Good grief....

Don't jump through Hooper hoops, please.

 

[krabapple]

No, I do not make it sound like he is not a member here. I am saying you hardly have a legal right to ask him to send the box for test to you,

What a weird thing to propose. No one has suggested as much.

 

[krabapple]

Audacity is very bad at rendering the plot from the calculated data. Making the window wide may improve things a little:

Why so strange plots in Audacity

This is a common stepped sine sweep from Ultimate Analog test record form Analog productions, track A6, recorded with RIAA active Why does the plot look normal in CoolEditPro2=Adobe audition but ro ragged and sloped in Audacity-Right side. What am I doing wrong, many people use Audacity but I...

www.audiosciencereview.com

But the underlying data seems correct. If you export it and plot externally you get:
View attachment 549622

That's comforting. It's what I traditionally do -- export the windowed Audacity data -- a column of frequencies and a column of levels -- then plot it in Excel. The plot images Audacity itself generates, I find hard to work with, and generally don't.

 

[krabapple]

You could have stopped there. Best to just avoid Audacity altogether.

Um, no. It doesn't claim to be a DAW, but it's perfectly fine for basic audio manipulation, including recording. And it's free. I'm happy to use it.

 

[krabapple]

Could not we stop it? We have no reliable valid data, so the results are untrustworthy, and nothing during those 49 pages has changed it. Period.

Sure, but I'd also suggest mods fork off the entire, now quite long, sub-discussion about oscilliscopes and testing gear and polarity, so we have a clearer narrative through-line here.

 

[JustAnandaDourEyedDude]

If Amir were to additionally pay the SPCA donation, how would he in turn "break even" when he wants to sell the comparator after he's done testing it? Particularly if he found a problem with its function? Find a buyer willing to pay over twice the going price? If the item is found faulty, perhaps an audio museum of the future would pay a high price for it, to display it in their "ways testing may go wrong" section?

Note: No non-human animals were harmed by the posts in this thread. Well, with the exception of quite a few golfing panthers, who have been pained to overhear gossip that human hearing is superior to panther hearing of DACs, and some scurrilous whispers that perhaps they should have been listening more attentively to the DACs instead of sneaking off to the golf course. They simply can't believe what they are hearing. Some human animals were indeed offended by some thread posts, but alas they have no prospect of being rescued by the SPCA.

 

[pma]

What a weird thing to propose. No one has suggested as much.

It might be a different language feeling - I am not a native speaker. However, my opinion is that it is in the competence of the box owner and he cannot be pushed to send it for testing. But we may ask for valid and reliable recorded data, otherwise the result of the experiment is useless. I would strongly recommend to participants not to lose their temper.
Enough, no personalities and let's stay technical.