And what makes you believe these measurement results should be taken for serious? I learned: Doing measurements is not easy at all and needs a lot of experience.

RME also had great problems with the "babyface"
wich is the worst interface ever in this test!

And now this ?
Strange !

/Göran Nilsson

I paid 1800 EURO for my ADI 2 pro AE (I am waiting for it)
but I paid less then 200 EURO for the Mackie.

And the Mackie is better so of course you can make a good
capture and a good mix with a 'modest' (Mackie) converter.

Also remember:
there is no sound differences between converters in double blind tests !

But if you use it for mastering with hardware effects and several passes
the "ADI 2 pro will" degrade the mix (that is just what this test shows).
And this is of course audible.

/Göran Nilsson

I did not dig very deeply into this test method, but what I did find out with some experimenting is how disproportionately large the influence of even minute phase shift turns out to be for the resulting "difference" - even if the files as such are otherwise absolutely identical. Depending on the sample rate, it can be very difficult to align two wave forms well enough to lead to maximum cancellation, even "half a sample" can make a significant difference to the value of this "measurement". My personal conclusion is that this test is simply no way to realiably determine a converter's audio quality. Returning the ADI-2 simply because of this figure may be a mistake. Trust you ears - and our published specifications, which are the ones that count, and whose results will not be determined by otherwise irrelevant factors like small phase shifts between DA and AD conversion, which may lead to a result of questionable value...

I wonder why no one has ever tried this test by feeding converter A's DA into the AD of converter B and vice versa, and then comparing - and wondering what the hell that result actually means... :-P

Regards
Daniel Fuchs
RME

To me that tests look interesting. I'm not in position to technically debate on their quality. I can't be sure one don't have some agenda behind it, nor the opposite. But I was interested in reading the comments, some are very reasonable on both sides. What that tests say is that for that particular method of testing some do better, some do worse. That's it and nothing more. I would feel fine about other methods of testing too.

And I would not be blindly angry at them if my more expensive hardware do worse than some cheaper ones in their particular test. That can be many reasons for that, like, the test technologies involved, human mistake factor, other variables (wires, connectors, electricity, grounding, etc.), some agenda, so on, and finally, poor engineering and cheap components behind the brand name (no implications here).

Also the feature they test is not the topmost important one for many people, including myself. Although for many it is.

I would call it "exactness" maybe, lol?

Yes, you are definitely right about timing. It means it is really hard to align with both recording and later comparison. Maybe some tests are just more "lucky" in this regards. It would help to do the sampling at a highest possible rates. Also would be great to implement some sub-sample delay in analog connection for several passes. Crazy.

Again, I'm speaking out of my mind thinking about reverse polarity null comparison, I did not dive deep into how their method works, actually don't want really, prefer others to do so.

Yes frequency nonlinearity and phase shift seem to be the main cause.
And it is both hard and expensive to get these filters really REALLY good ? right ??

The test measures how well one pass of loopBack preserves the "mix".
In a mastering situation you may have several of these passes.

Before Mr Didier Brest started his maraton series of never ending
acumulation of rather dubius data there was this AMS who made several passes and
there it was a clear audiable difference between interfaces (high end better).
Gearslutz deleted AMS. (I hope I remember everything right now.)

Mr Brest now don't use The Diffmaker software (which was presented on an AES meeting)
but does his own calculation in Mathlab. But he presents the values from diffmaker also.
His method is not scientifically acknowledged as I understand it.

I will probably keep my ADI 2 pro AE (very good reviews from everywhere).

There are many  "A's DA into the AD of converter B" result in that infinite list.
From high end DACs To high end ADs with very very good results.

For example:----------------------------------------------------------------------
MOTU 828mkII ---> RME Fireface 400, 2 tests: MOTU or RME clock (MusicManic)
0,3 dB (L), 0,5 dB (R) Corr Depth: 24,5 dB (L), 26,3 dB (R) Difference: -44.1 dBFS (L), -45.3dBFS (R)

and the other way

RME Fireface 400 ---> MOTU 828mkII, 2 tests: RME or MOTU clock (MusicManic)
5,1 dB (L), 5,2 dB (R) Corr Depth: 35,1 dB (L), 36,7 dB (R) Difference: -54.7 dBFS (L), -55.7dBFS (R)

Maybee the RME AD is better than the RME DAC here?--------------------------------------

There is really no audiable differenses between converters. But yes you can hear the different
filters (very little). I guess that RME's problem is the DAC-filter (the CD player brickwall filter).
People also say that RME sounds clear in a harsh way.

/Göran Nilsson

New test results for the ADI 2 pro is coming out.

Filters:
Slow_In  Slow_Out

ADM: -1sec, 0,351dB (L), 0,344dB (R)..Corr Depth: 33,6 dB (L), 35,2 dB (R) Difference: -52.4 dBFS (L), -53.5 dBFS (R)
Matlab: -1.000113818 s 0.3521 dB (L), 0.3453 dB (R) Difference: -52.8 dBFS (L), -54.3 dBFS (R)

These are acceptable numbers for the 1600 EURO price.

Old numbers not so good:
SD_Sharp_In  SD_Sharp_Out
ADM: -1sec, 0,366dB (L), 0,357dB (R)..Corr Depth: 33,6 dB (L), 35,3 dB (R) Difference: -43.2 dBFS (L), -44.6 dBFS (R)
Matlab: -1.000125149 s, 0.3521 dB (L), 0.3504 dB (R) Difference: -44.5 dBFS (L), -45.5 dBFS (R)

Everything now indicate that RME has to just tweek their slow filters to get still better results.

/Göran Nilsson

I will listen to the AD converter and compare it to my other interfaces.
In fact I can do that (most people can not do that because they don't have that kind of setup).

But I don't have any means to compare DACs. And I don't bother so much about DACs. I dont even know
how that should be done. I have good AudioVector speakers in a treated room so that part is OK
but no way to change DAC during listning.

And yes (people say) "'the ADI 2 pro' has excellent sound quality plus a little bit more resolution"
and is not as harsh as the other RMEs.

/Göran Nilsson

Yes that should be possible here too.

But how to set the loudness the same and
within 0.1 dBC ?

/Göran Nilsson

I mean the SPL level from the Speakers must be the same when you switch DAC.

/Göran

Yes, its doable by ear comparison.

MC you're right! I've mentioned it many times in that topic: The test IS flawed, users without a clue of how to gainstage are randomly recording one cards input in another cards output. Then Didier "calculates"the "difference" and then "corrects"it with mathlab

why this test is flawed?

1. different input and output latency
2. different signal levesl
3. different RFI enviroments (desktop /laptop /USB PSU/Battery, ground no ground etc)
4. random outputs "recorded" on other gears input

If you want to do anything meaningful you're better of calibrating your IO loop in REW or use an RTA with an impulse

(Post edited for correct quote levels - RME Support)

I am not a specialist with Diffmaker nor have I ever tried it personally (when I wanted to do that years ago it just kept crashing on start-up already), so I have to rely on the available information found in the GS thread.

Diffmaker is said to do a null test between the original file and the newly recorded file. This nulling is done automatically and with sub-sample accuracy, otherwise it woudn't work at all (at 44.1 the chances are like zero to have both files sample-accurate).

Diffmaker's results have people wondering since years what is wrong with it, as its over-exagerated results seem to have an internal weighting that often contradicts other measurement methods.

Maybe the filters of the ADI-2 Pro now easily explain it all. When Diffmaker is doing the nulling, parts of the file are still not nulled - because any analog and digital signal path includes phase shift, at the lower end and the higher end. This phase shift is inaudible, a proven fact.

With the ADI-2 Pro you have phase-linear and minimum phase filters at your disposal, on both AD and DA side freely selectable. Now we see that the most transparent and less damaging filters SD Sharp (which have said phase shift) deliver bad results, while the intransparent, treble reducing and partly aliasing causing linear-phase filters are 'measured' as superior. These results indicate that Diffmaker ist completely centered on the full nulling with a weighting that is unusable in real-world, and will even indicate the opposite of what is true.

The penny hasn't dropped it seems, so I give you a simple example:

Do exactly that. Connect input to output at same ref levels with everything EQ etc turned off, then re-record one file 10 times in a loop. A 10 second sample is enough. Use SD Sharp on AD and DA. I guarantee that you will have a very hard time to hear a difference between the original and the 10 pass file, if you manage that at all.

Please note that intentionally the DA has a bit lower level than the AD, to prevent overloads in loopback applications. The difference is small (around 0.1 dB), but after 10 times it will be more than one dB, so you need to normalize or manually correct the level of the 10 pass file to be able to reliably listen for sound differences.

Now do the same test with the Slow or NOS filters (phase linear). Treble is lost already on the first pass! After 10 passes you get a dull sound, all glitter removed.

Thinking about it that would also be a perfect reason for the developers of this project to abandon it. Nulling can't work with analog/digital conversion and all the filters that are needed and unavoidable.

It also opens the way for another theory: units that show extreme high Diffmaker scores are lucky to have the DA phase shift be compensated by an inverse AD phase shift. They will get super scores for something inaudible, while the real sound quality based on SNR, THD and frequency response is not, or only a little involved in that score.

Hmmm, I spent too much time with this stuff again...

The "Gearslutz List" has been updated with several diffrent filter combinations:

RME ADI-2 Pro, balanced out to balanced in (LesC)
Sharp DA - Sharp AD: 0.4 dB (L), 0.4 dB (R)..Corr Depth: 33,6 dB (L), 35,2 dB (R) Difference*: -53.9 dBFS (L), -55.2 dBFS (R)
Slow DA - Slow AD: 0.4 dB (L), 0.3 dB (R)..Corr Depth: 33,6 dB (L), 35,2 dB (R) Difference: -52.8 dBFS (L), -54.3 dBFS (R)
SD Sharp DA - SD Sharp AD: 0.4 dB (L), 0.4 dB (R)..Corr Depth: 33,6 dB (L), 35,3 dB (R) Difference*: -44.5 dBFS (L), -45.5 dBFS (R)
SD Slow DA - SD slow AD: 0.3 dB (L), 0.3 dB (R)..Corr Depth: 33,6 dB (L), 35,2 dB (R) Difference*: -47.9 dBFS (L), -48.9 dBFS (R)

Sharp DA - Sharp AD :  -53.9 dBFS (L), -55.2 dBFS (R)

Are really GOOD ! Well Done RME !
The ADI 2 Pro is also the best of RME's interfaces.

/Göran Nilsson

Yes, we are wasting our time - again. Instead of trying to understand what I wrote in post 28, or even to do the suggested test, the content of my post is fully ignored. Of course one might say, as it would cause the Diffmaker believers world to collaps. Still disappointing.

I think we all have to (at least try to)  keep some concepts apart here.

1.)
The "GSloopBackTest" do not measure audio quality. No is does not. I totally agree with Mr MC.

BUT

2.)
The "GSloopBackTest" testresult is correlated with public perceived sound quality.

---------------------------------
Correlation does not imply causation:
https://en.wikipedia.org/wiki/Correlati … _causation       

/Göran Nilsson

If the Diffmaker results/figures touted as very good were SNR values or the like, they would be absolutely terrible...

I could post the following on GS, but I'd rather post it here, it will possibly find its way there.
I had an idea how to test the test, slightly different from MC's suggestions. There is no conversion involved, this uses digitally created files only. I hope I did not make any systemic or other mistake here. If so, I'll gladly accept any correction. Feel free to try this with other parameters. I don't know whether any such thing has been tested in all the years since this has been going on.

I created a 384 kHz Sequoia project out of which I exported 44.1 kHz files of identical length, but with different audio content (at the best possible SRC quality). One set of files with just a 1000 Hz tone, one with a combination of 100 and 1000 Hz tones. I adjusted levels and the position of the audio objects within the project, and exported the identical range to create files of the exact same length. I used the best possible values for time alignment and level adjustment in Diffmaker while comparing files.

First test: Comparing identical files, the result is, as should be expected:
Parameters: 0sec, 0,000dB (L),  0,000dB (R)..Corr Depth: 300,0 dB (L), 300,0 dB (R)

1000 Hz file tests, files with following changes compared to the original:

• Level changed by 0.5: 0sec, 0,556dB (L),  0,556dB (R)..Corr Depth: 182,7 dB (L), 183,4 dB (R)

• Level unchanged, position of audio object shifted by 1 sample (@384), to simulate a subsample delay @44 during conversion:
  (INV)parameters: 497,4usec, 0,000dB (L),  0,000dB (R)..Corr Depth: 84,3 dB (L), 84,3 dB (R) (not sure what INV is for, could not find anything).

• Level unchanged, object shifted by 10 samples:
  Parameters: 974usec, 0,000dB (L),  0,000dB (R)..Corr Depth: 92,5 dB (L), 92,5 dB (R)

Identical results for files that were shifted and level-changed.

By Diffmaker standards applied to converters, these would be great results, but shouldn't  level differences and esp. time offset be compensated even better?

Second test, now it gets more interesting and seems to confirm what MC suggested (phase shift):

Combination of 100 and 1000 Hz tones, files with following changes compared to the original:

• Identical files: 300

• Level changed by 0.5 dB: Parameters: 0sec, 0,501dB (L),  0,501dB (R)..Corr Depth: 181,9 dB (L), 178,4 dB (R)

• Level unchanged, 100 Hz object shifted by 1 sample @384
  (relative to the position of the 1000 Hz object):
  Parameters: -229,8nsec, 0,000dB (L),  0,000dB (R)..Corr Depth: 58,2 dB (L), 58,2 dB (R)

• Level unchanged, 100 Hz object shifted by 3 samples @384:
  Parameters: -9,421usec, 0,000dB (L),  0,000dB (R)..Corr Depth: 41,3 dB (L), 41,3 dB (R)

• Level unchanged, 100 Hz object shifted by 10 samples @384:
  Parameters: -2,298usec, 0,001dB (L),  0,001dB (R)..Corr Depth: 38,2 dB (L), 38,2 dB (R)

Identical results for files that had shift and level-change.

This is not complex musical material with a 20-20k range going through DA and AD conversion, these are digitally created files of a rather simple nature. It should be clear that there can be no audible difference between these files with such miniscule changes. Yet, the difference shown in Diffmaker results is quite significant and would suggest differences in "audio quality" that don't really exist.

Again, I hope there is no error in this test... The simulated "phase shift" values I used here are random, I don't know whether they would be in any way similar to real world values of DA/AD conversion. The idea was just to try to demonstrate how inaudible differences could cause results in Diffmaker that create a different impression.

Regards
Daniel Fuchs
RME

I have now sent my "ADI 2 pro AE" back.

I compared it to my Steinberg UR824 and there was
not much (if any at all) of an improvement.

(I tried to be as objective when listening as possible.
Same loudness, treated room and good speakers. This can
be rather boring but is of course necessary.)

I will look for something more "HIGH END".

(I have really listened to it (both AD & DAC) and try to
like it but it was not up to my expectations. I am so sorry.
But I have learned much from doing so and that is the most
important thing for me.)

/Göran Nilsson

You are more likely limited by the capabilities of your speakers. I have heard the ADI-2 Pro and DAC in front of an ~€30k HiFi from my friend. Accuphase E600 with builtin Accuphase DAC-40 card and B&W 803D3. First it was hard to identify a difference, but after longer comparisons I tend to say that the ADI-2 Pro/DAC delivered a better more authentic presentation of the room where i.e. the vocalist was slightly better separated from the rest of the instruments in the background. For comparisons like this you also need to choose your sound material wisely, that the mix has good quality and you know this material well.

And even for my own use cases, i.e. in front of Geithain RL906D and with an Audeze LCD-3 phones it sounds better than before, although the upgrade from an RME UFX to UFX+ was nice already.

So the product is without any doubt on par with DACs being used in the high-end area and in regards to the additional features that RME bundles with it even better than many other products.

So if you say now that a low price product sounds the same as the ADI-2 Pro then its clear to me that either your speakers or your ears have deficits, so that its at the end not possible for you to recognize a difference between the low cost product and the ADI-2 Pro.

IHMO it doesn't seem to make any sense for you to spend even more money for another DAC with "questionable" more quality, but definitively less features.

Sorry, but from what I read from you I think your biggest problem is a certain lack of know how, experience and by this you come to wrong conclusions.

But maybe a new DAC will sound better to you because of psychoacoustic effects, if you pay more your brain will have the pre bias that it will sound better and then it will sound better for your ears. Brain can fool you very easy.

Good luck.

Obviously your monitoring is pretty bad otherwise you would have noticed the difference immediately. ADI-2 Pro is sounding much cleaner with deeper low end and less distortion than anything I heard. It avoids the jitter which is pretty aparent at slightly louder levels, as you don't get the harsh spikes that so many converters (even high end) produce. Sound is smoother, but transients remain intact. Actually unbelievable clocking and conversion + all other features for the price. It makes competing product appear what they are - overpriced.