Why are digital streaming equipment manufacturers refusing to answer me?


I have performed double blind tests with the most highly regarded brands of streamers and some hifi switches. None have made any difference to my system on files saved locally. I have asked the following question to the makers of such systems and almost all have responded with marketing nonsense. 
My system uses fiber optic cables. These go all the way to the dac (MSB). Thus no emi or rfi is arriving at the dac. On top of this, MSB allows me to check if I receive bit perfection files or not. I do. 
So I claim that: if your dac receives a bit perfect signal and it is connected via fiber optic, anything prior to the conversion to fiber optic (streamers, switches, their power supplies, cables etc) make absolutely no difference. Your signal can’t be improved by any of these expensive pieces of equipment. 
If anyone can help explain why this is incorrect I would greatly appreciate it. Dac makers mostly agree, makers of streamers have told me scientific things such as “our other customers can hear the difference” (after extensive double blind testing has resulted to no difference being perceived) and my favorite “bit perfect doesn’t exist, when you hear our equipment tou forget about electronics and love the music”!
mihalis
Its audio not rocket science. If you want bit perfect you can have bit perfect and you can do it very inexpensively.
Sure you can. But how is the "bit-perfect" data being translated and rendered?

If I am using a player software with digital EQ and a DAC or interface like the MSB still shows that it's "bit-perfect" data, shouldn't that tell you something?

As a visual analogy, if I switch between color space outputs on my video streaming device to send a different color gamut to a capable display over HDMI, the signal to the display is still "bit perfect". Yet, if the display is capable, it will result in a different color space. If I set the wrong color space for the source material, the image(s) will end up altered.
Its audio not rocket science. If you want bit perfect you can have bit perfect and you can do it very inexpensively.
There are a lot of great points made here on both sides of the argument but I think as with anything many people are looking at things too microscopically and need to zoom out and take a macro view at what is actually happening with "streaming" on a network.

One thing no one seems to bring up in these back-and forths is the essential architecture of a network which consists of layers. Sometimes you might hear about these layers in jargon like "stack" "full stack", or other such lingo.

If one happens to be a competent "full stack" software engineer, then the challenges of figuring out whether or not the data are/is good is largely irrelevant.

What audiophiles fail to realize is that vast amounts of opportunity for mishandling/misinterpretation of "1's and 0's" can, and often do happen at the final few stages in the process (presentation, application) of translating the "digital" signal information into meaningful use by your device.

The fact is that not all "streamers", let alone music player software, are created equal, and there are many poor ways of going about it in fact.

From a technical standpoint, the data received by a streamer from a network is the exact same data another streamer can receive on a network.

Where the conversation gets tricky is what you are using to render the data and how it handles the various software processes to de-code the information (1's and 0's).

As an analogy, gamers spend varying amounts of dough on better graphics processors. No computer engineer would argue that the raw game data being received by two different GPUs is different; you can perform a hash/checksum to verify the data is all there.

Equally, no one will argue that two different GPUs will and can produce different results when finally rendered to your monitor (not to mention the monitor has it's own internal processing to deal with to receive the data).

What's so funny to me about "science only" audiophiles is they don't tend to think about the actual science much.

Thank you cal3713, I do remember that paper ages ago. I will point directly to the conclusion:

Our main conclusion, however, is that it is not necessary to invoke any advantage for hedonic judgments to explain our earlier results. These, and the new results here, are just particular instances of the advantage in statistical power that Ennis(1990) shows forced-choice methods to have over triangular tests. More consistent judgments are a consequence not of greater sensitivity to hedonic differences but to the statistical properties of the decision rules followed in different tasks.

I believe I noted above that ABX was statistically more robust.
@audio2design  Below is a journal article you might find interesting.  It explores a mechanism behind the empirical paradox that people can show a reliable preference between two stimuli but fail to discrimination between them on an ABX discrimination test (here referred to as triangle testing). 

I will note that the reason they identify in this case is actually "the statistical properties of the decision rules followed in different tasks."  I still suspect that raw preference judgments are more sensitive than discrimination judgments but that was not the driving factor for differences in this case.

https://link.springer.com/article/10.3758/BF03205304




audio2design
And there is it folks, can’t win an argument based on facts, or truth, so must resort to an insult and labelling.
Nope, I offered no insult. There’s no need for you to pretend you’re insulted simply because someone disagrees with you.
Hate to break it to you "cleeds", but ABX has absolutely nothing to do with measurement.
That’s correct, of course. The only function of an ABX test is to determine if the listener - under the test conditions - can reliably distinguish whether "X" is either "A" or "B".
... in another thread, you used as an argument, Wireworld coming up with a "Patented" device to improve their ability to do double blindtesting. Cake and eat it too? Do please try to be consistent.
No, I never, ever said that. Ever. Please try to be accurate when you make claims. It’s bad enough when you use illogic to make an argument, but it’s worse when you fabricate claims made by others.

I have pointed out that Wireworld offers a comparator to use in evaluating cables. It’s a little odd that you seem to insist that others do this experimenting for you, but I understand that if you were to actually conduct such testing you fear the results might conflict with your fundamentalist beliefs.
Lot’s of people have lots of "beliefs" ... Does not change the lack of evidence that ABX works and hence is the "gold standard" or as close as we have to one.
Yes, and your measurementalist "belief" is that ABX is the gold standard for audio testing. As I’ve noted many times previously, it’s a very useful tool, and I’ve been an ABX test subject and found the results v-e-r-y interesting. It absolutely has a place in audio testing.

But ABX testing is just a tool - a single, solitary tool. It is not a path to Absolute Truth. That apparently offends your belief system.

Taken to the extreme, there are those who actually believe  "if you didn't hear it blind, you didn't hear it." Obviously, many sighted people can hear just fine.
And there is it folks, can't win an argument based on facts, or truth, so must resort to an insult and labelling.

Not everyone shares your measurementalist’s belief that ABX is the "gold standard"

Hate to break it to you "cleeds", but ABX has absolutely nothing to do with measurement. I think that is why it is so threatening, because it is not measurement based. It is based on one thing, and one thing only, human listening. Your ears and brain. No scopes, no meters, no Audio Precision equipment, just some stereo equipment and your ears.


What's funny is in another thread, you used as an argument, Wireworld coming up with a "Patented" device to improve their ability to do double blind testing. Cake and eat it too?  Do please try to be consistent.

Lot's of people have lots of "beliefs". Lots of people don't have the educational or practical experience either. Does not change the lack of evidence that ABX works and hence is the "gold standard" or as close as we have to one.
audio2design
... double blind ABX testing is considered the gold standard ...
Not everyone shares your measurementalist’s belief that ABX is the "gold standard" for evaluating audio equipment. Not even close. It appears that really upsets you.

That doesn’t mean that ABX is useless, of course. But it’s just a tool - a single, solitary tool.
And it is quite clear that you will continue to misunderstand and misstate the processes involved in an ABX audio test, even though you admitted you really don't know what happens.  But, nice to state, effectively that "we'll" never know because "you" don't run an audio testing laboratory. I am sure no one who does this style of test has any experience in testing human perception ....

It's rather "interesting" that "memory" or pathways, or whatever, are good enough to "remember" well enough to know if they prefer A or B, but not well enough to remember if they prefer C more than A, or C more than B.  That is what you are stating even if you think you are not. 


You whole argument is based on assuming a process that you admittedly don't know, and then assuming it must not be the one you feel it should be or would be more successful. That is bad science.


As a counterpoint, people who are not "audiophiles" have been shown, several times, to be more adept at detecting minor differences when they are trained, i.e. taught what the differences are likely to be, and given examples. They create the appropriate pathways for detection of differences.  To be clear, more adept than "audiophiles".  Based on communication of audiophiles when they compare cables, certainly on here, I would say that preference is exactly the method they use, or at least claim to.

Here is the thing. When comparing two of anything in audio, like in AB or ABX testing, the descriptors are invariably related to preference. More natural. Improved soundstage. Pinpoint imaging. Tighter bass. Sweet mids. These are comparative descriptors, not unary descriptors. That indicates preference.


Let's not forget that while double blind ABX testing is considered the gold standard, there is no more success achieved in AB testing either, which as per the used descriptors, are preference related.
When people get the X test with audio equipment I suspect many are trying to figure out if the treble, bass, image depth, tonality, etc. match A or B, not asking themselves how much they like the presentation and then seeing if that preference is closer to how they felt when they listened to A or B. Of course we’ll never know if that’s true because unfortunately I don’t run an audio testing laboratory.

And yes, there is learning happening in the memory patients, but the area of the brain that makes explicit identification judgments does not have access to it. I’ll also note that when they fail to recognize a previous acquaintance that new/old judgment is even easier than matching to a particular object as required to succeed in ABx. Despite utter failure of explicit identification, the preference system chugs along just fine, leading to adaptive decision making.

This dissociation is why it is incorrect for you to state that, "if you can’t match A or B to X, then you can’t tell the two apart and you hence have no preference as you don’t as actually prefer either."

Anyway, it’s clear you’re going to continue to believe that matching judgments are the appropriate way to do A B testing. And that’s 100% not how I would do it if I were optimizing people’s decision making. It doesn't matter if you can pick your stereo equipment out of a lineup, it matters whether or not you like it. So be it...
Your example is an absolute identification test. As noted this is not applicable to the discussion. You seem to be missing the point.  If you have AB and X. And have a definite "preference" for A, then when x=A, that preference should replicate.  You try wine A and B side by side. You claim you prefer A to B.  Now I give you wine X. Do you claim you prefer it to wine B? If so, it must be A right? What if it is actually B?  That means your "preference" was random.

They have studies with patients who have zero long term memory. Every day they fail an X test by failing to identify people with whom they've interacted with repeatedly.

  I suspect if you instructed people to base their identification judgments in an abx solely on preference they'd do significantly better.


Actually what results in better ability is training in the characteristics of differences. The natural tendency is to rely on "preference", which is very fickle.

Nonetheless, they form adaptive preferences for these individuals based on whether those past interactions have been positive or negative.


Which would require learned neural patterns. This is not related to directly to ABx and more related to why blind testing is necessary to ensure the learned neural patterns for looks are removed from sonic decisions. However as applied to ABx, those learned neural patterns should trigger the same for A and for A=x, as opposed to B.  It is also why training improves ABx as you develop additional neural pathways for characteristic detection which is what preference is. 


And just to highlight. With those memory patients, the preference test is highly reliable (defined in testing science as a consistent, repeatable judgment). Their performance on the identification test is completely unreliability (i.e., no different than chance). You do not need to be able to say x = a to make a reliable preference judgment.  I suspect if you instructed people to base their identification judgments in an abx solely on preference they'd do significantly better.
It's about saying a = x. That is an identification judgment. You don't have to believe me, but the brain is worse at that than at determining preference. 

They have studies with patients who have zero long term memory. Every day they fail an X test by failing to identify people with whom they've interacted with repeatedly.

Nonetheless, they form adaptive preferences for these individuals based on whether those past interactions have been positive or negative.  That is because the area of the brain responsible for preference is different (and more basic & important) than the area responsive for identification.
Again, you obviously don’t understand ABX testing which is NOT about absolute identification, i.e. picking out the 98 Bordeaux from the 96 and the 99, it’s about having 3 bottles, two 98, and one 96 and being able to tell the two 98 are the same. If you can’t tell a 96 and 98 apart then you don’t actually prefer one over the other. ABx increases the statistical reliability of an AB test. ABx testing IS a preference test essentially as it requires no absolute identification.
Please tell me how matching is different than identification.

And please see above for a discussion of why the brain is better at determining preference than identification.

https://scholar.google.com/citations?user=NQGcNPQAAAAJ&hl=en
If you were a PhD with said qualifications, you would know the X has nothing to do with identification, it is to test reliability. If you can't match A or B to X, then you can't tell the two apart and you hence have no preference as you don't as actually prefer either.


Given the light discussion around ab(x) testing, as a scientist with phd training in human data collection and cognitive functioning, I object to the (x) part of the judgment task.

Identification should not be the goal. We listen to equipment to decide which we prefer... which is better. That should be the decision when doing (blind) ab testing.

Given the light discussion around ab(x) testing, as a scientist with phd training in human data collection and cognitive functioning, I object to the (x) part of the judgment task.

Identification should not be the goal. We listen to equipment to decide which we prefer... which is better.  That should be the decision when doing (blind) ab testing.

If I were designing the test for audio equipment, I'd let people switch back and forth as many times as they like, taking as long as they like for each (level matched) sample.  Eventually they just decide which they like best. Repeat that task over and over throughout the course of listening to music and I guarantee that people will begin to find their true preference and that that preference will be consistent. 

It's very easy for the brain to develop and make decisions based on good/bad/preference and much harder to make the decision based on label identification. They are different processes in the brain and preference is the more basic, a primary system. Newborns know what they like and what they don't and will often show good a/b consistency. The brain doesn't even need to identify a stimulus to make this judgment. That comes later because it's less important. First, figure out if something's good or bad, then if you've got time, figure out what it is.  This is the order of processing for all of us.

The eye doctor uses this ab-preference testing when they're figuring out your prescription. They flip between two possible magnification factors and you just tell them which is better. It works great for letting you find your own way to an optimal solution. They don't care about labels or giving you mystery options and having you say whether it corresponds to a or b because that's irrelevant. It would make the process less effective. 

And finally, I'll just say that if you don't believe in blind testing, you're lying to yourself. There's decades of research demonstrating the impossibility of avoiding the biasing effects of pre-existing knowledge. People have used samples of scientists trained in this area of research and they are just as influenced as you or I. You are not immune.

That said, it's hard as hell to do blind testing and I never do. Fortunately placebo effects are very real and they help your preconceived decisions feel right even if they were wrong.
Jmphotography, in the system you describe there is no fiber optic blocking all noise and as a result I completely understand that there could be improvement with a good streamer. Thank you for that feedback.
The Supra was designed by Toyota, using a bevy of BMW parts from the Z4, and ASSEMBLED by Magna Steyr. This is a contract manufacturing model that probably applies at least at some level, to much of the electronics industry at some level. Even Apple does not see value in setting up their own manufacturing lines.

That is much different from the model you are using with Foxconn where Foxconn sells it for rebadging. This is not contract manufacturing it is ODM (original design manufacturing). It does apply to some televisions, usually lower end models like Magnavox, RCA, etc., and perhaps some low-mid end models in other manufacturers, like Philips, but not to most of what you buy from say Samsung, LG, Panasonic.


I am not aware of any streamers in the audiophile world that would fall into the ODM category. That would be low end offshore DACs.
One reason might be that the distributor knows nothing, because some company like Foxconn designed and built it, and sells it to them to market.  Have you ever wondered why so many companies' televisions have stick on brand names.
It is not just electronics, either.  The new Supra is said to be made by BMW.  It is not, and neither is that BMW.  An Austrian company that designs and makes about 200,000 cars per year makes both, as well as the Mercedes G series, and others.  Like Foxconn with electronics, they design and make stuff, then sell it to other companies to provide drive trains, and then to market.  You can verify this by entering the VIN into your computer.  Also, the Toyota Sports car, that is the same car as the Subaru, has a VIN beginning with JF, for Subaru's  parent company, FUJI. 
German built cars built by BMW begin with WB.   W is for West, as in West "Germany".   Mercedes is WD (Daimler).  For vehicles made under contract for Mercedes,  ZA and KN are used.  If it starts with a J it is made in Japan, with numbers 1-5 in North America (USA 1-3).  The first letter of number is the country code. Who know who makes what anymore?
Its incorrect because its wrong. Your whole premise is false................Not only wires, I can tell the difference between lots of things- wire on the floor vs elevated, wire going one way vs another, warmed up vs cold, on springs vs cones, on and on. And on.
LOL.  Thanks for the facts mihalis and welcher.  Great discussion.
Since I am receiving bit perfect signals nothing happening prior to the conversion to fiber matters IMO.
mihalis, that's the beauty of it.  With my WiFi computer noise, speed, amount of RAM etc. don't matter, but there is hardware prior to fiber and it does matter.  As long as you use the same hardware with different streaming providers result should be the same.
I’ve auditioned Node2i, raspberry pi4, SimAudio, Bel Canto, Innuos, Lumin, Auralic and I haven’t noticed a dimes worth of difference between them when used with Roon. I will say I have noticed improved sound through a couple  when using their APP but I attribute than to differences in the software.
My test say different. I just got into streaming back in the summer and purchased a Bluesound Node2i. While it was a nice a streamer for the money, I was missing something and I also wanted a server to rip disc to replace my old Sony DVP9000ES. I bought an Innuos Zen mkIII. Yes almost 5x the cost of the Node 2i. I received it this past Monday and hooked it up. My daughter who is now 22 and has started appreciating music but is no audiophile like me, she was sitting on the sofa on her phone. I put on a familiar song and her head perked up as did mine. The Innuos was much fuller, richer in sound. Once you sat in the magic spot you could hear the improved soundstage. We let it play a while and then switched back to the Node since I still had it hooked up to the same Qutest DAC, we both looked at each other and smiled. Even my wife heard a difference. My system is pretty simple. Older pair of B&W CDM9NT being fed by a PrimaLuna EVO300 integrated being fed by a Chord Qutest DAC previously being fed by the Node, now the Innuos Zen.
Getting “bit perfect” signal is not a rocket science. You make it sound like people are all naive and streaming music from some noisy desktop computers. It’s extremely easy with all major streaming protocols, including from Roon via a NUC, Nucleus, Innuos, and so forth.

MSB Select DAC starts at $85,000 I believe. Once you add your ProISL / ProUSB module (that alone is $2,000) and perhaps the second power base, and the upgrade clock, then the price becomes ridiculous.
Thyname it’s the select dac with the same interface you describe. Since I am receiving bit perfect signals nothing happening prior to the conversion to fiber matters IMO. If the signal was not bit perfect (msb offers an extensive test) then it could have affected things. 
The test you did Mihalis sounds well thought out and the results exactly as expected.

It is possible for the incoming data stream to impact clock timing but that would be expected in a poorly constructed $50 USB DAC from China not an MSB, not to mention the pro-ISL fiber does not communicate USB.


There is also no guarantee that thyname's high end streamer does not manipulate the bit stream or that what he compared it against manipulated the bit stream and the one he prefers does not. We like magic and so we hold on to the magical explanation when the more likely one is almost always right.


Keep in mind a lot of high end components are intentionally euphonic and their euphonic signature dominates their sound not necessarily their technical competence.
@mihalis I simply shared what I experienced, with that MSB DAC and that same module. Which MSB DAC did you own? I don’t see it under the System for your current gear. So for the ProISL / ProUSB module combo, was it the same you had? That’s what I had.

My subsequent attempt in technicality of the benefits of USB protocol in streaming (or lack thereof) was simply based on my limited technical knowledge. I don’t pretend to be fully able to explain everything I hear.

As for the MSB ProISL / ProISL implementation, yes, the main objective is to eliminate copper, therefore the electrical interference associated with it, by introducing fiber. But it’s not as simple as that. The first device, what they call ProUSB (the dongle that stays outside the DAC, on it’s own) converts USB to their proprietary ISL audio transmission via the fiber. Then the module that goes into the DAC, what they call the ProISL module, is the one that is directly connected to the DAC architecture. Basically copper USB cable to the ProUSB dongle input, then fiber optics cable from the ProUSB dongle output, to the ProISL module (attached to the DAC) input.

So it’s not as simple as fiber, and here you go, all problems solved magically. The conversion itself is perhaps just as important, as in theory can introduce other issues.

Further, many of the non-believers will tell you that any of that is not an issue at all, and any competent Topping China $50 DAC has all that built in, galvanical isolation and all, so totally unnecessary to go through fiber they will tell you. So here is that.
Thyname thank you for the more extensive response which I hadn’t seen until now. Going for that bath now! M.
Thyname you wrote feeding via usb and I wanted to be sure that the prousb conversion was involved. So if you know more about how your streamer would improve the signal it would be great to hear it. It goes against what msb and other steamer makers are saying but maybe there is something they are missing. I haven’t had the opportunity to listen to this steamer, looks terrific.

kijanki much appreciated. Exactly, I am focused on the fiber connection because it eliminates all electrical interference. What I struggle with is that since the signal is bit perfect and the fiber eliminates noise which would have affected the dac, wouldn’t we be getting data = what was sent? Thanks again, getting late here in Europe. 
Some really good points about A B which I will address here. Since someone attacked my person for some mysterious reason, I am forced to explain that these tests were always done with a group of experienced audiophiles in Asia and in Europe. To them we added some friends who aren’t audiophiles to make sure we don’t have any bias from that. We did double blind tests, not triple blind. The groups were not the same always. And the tests were two hour affairs on occasion and not some extensive lab test. I am claiming nothing other than what this meant for us. 
The post earlier about how the ear isn’t made to compare is I believe correct but our methodology probably eliminated this bias. 
First we listened with knowledge of what equipment we are hearing. Then we listened to equipment which was given a code (not A or B but eg 346 and 589). And finally we listened to unidentified equipment in random order and asked to identify which code it was. 
Various pieces of equipment had different success because many obviously do sound different. But with equipment which shouldn’t really make a difference according to some logic, we had the following results

1) consistently, expensive equipment was identified as better and we could all elaborate that we heard specific differences. I certainly did think so

2) consistently, when the equipment was labeled, we could not necessarily say if it was overall better or not than the other but we developed certain characteristics we each thought we heard. These characteristics were not consistent between each listener but the same for each listener

3) not one person was able to identify these differences with more than 20% consistency vs 1 or 2!!!

Take me out of the equation since I am apparently a troll. The people in these rooms are audiophiles with significant experience, from different cultures and of different ages. We were stunned. (Non audiophiles usually abandoned us after 30m).

What this meant for us? If we smell something doesn’t make sense, we look into it so that we don’t end up spending our savings towards the wrong equipment or system. Ie we try to reduce bias. As a result we feel we have better systems with less equipment and less cost. 
This is how this worked for us. I am sure everyone has their way and I am not making any broad généralisations about anyone else. I did feel validation when I read the seminal work of o toole and some sites which do measure equipment extensively with correlation to performance potential. 
To each their own, although I hope at least some of you may consider these points.

troll out!
Hi @mihalis yes, it’s the ProUSB / ProISL combo:

https://www.msbtechnology.com/dacs/prousb/

It’s the two part combo (scroll down on that link above):

  • Pro ISL Input Module
  • Pro ISL cable
  • Pro USB
That’s what I wrote on my previous post, no? Is this not what you meant?

I don't think people realize that the USB signal does not get reclocked by the DAC. What they reclock on the DAC is the audio signal that is transported via USB.
In async USB DAC receives "frames" of data usually at 1kHz rate and feeds them into buffer.  It signals back upon buffer under/overflow to adjust size of incoming frames.  It is data coming in - not "audio signal".
I assume you meant prousb and not usb? Did you run bit perfect tests to ensure you were comparing apples to apples? 
It is encouraging to hear you got improvement and that it was pretty simple. Would you mind elaborating how you can intuitively figure out the reason based on what the innuos statement has / does? 
As for a USB streamer, I wanted to clarify the “clocking” part on this discussion, based on what I researched and experimented with back then (I no longer use USB now):

Yes, the AUDIO SIGNAL on the USB streamer is “clocked” by the DAC. No question. A good USB streamer reclock the USB signal, not to be confused with the clocking for the audio signal. It is reclocking the USB commnunication between the source and the DAC, meaning it does not work at the audio level but at the USB protocol level.

I don't think people realize that the USB signal does not get reclocked by the DAC. What they reclock on the DAC is the audio signal that is transported via USB. The USB communication components themselves (I.e in a USB streamer) are affecting the performance of the DAC. This has also independent of the galvanic isolation of the DAC, which on its own is a separate, independent issue.

I hope this makes sense @mihalis . Needless to say, theory is just that, theory, and not a substitute for practice.
@mihalis  yes, asynchronous USB receives just data and uses own independent clock for D/A conversion.  Still, any electrical connection can inject noise and alter this clock timing.  As for optical connection - it helps, but it is not a panacea.  Bits are still bits but timing of arrival can be altered.  Any system noise can make transition (including light intensity) "jagged" varying exact moment in time of level change recognition (threshold).  I don't know how to explain it better, but let's try this - Imagine in slow motion filing your tub with water.  It supposed to stop when crossing certain fill level, but water have waves (noise) and every time you repeat it - it stops a little bit sooner or later. That is time jitter.

Perhaps you know all this, but imagine that you receive 1 kHz pure sinewave.  Big electrical noise added to signal is causing different moment of level recognition and stream to "vibrate" in time.  When words fed to D/A converter vary in time then it will result in creation of additional signal - sideband frequencies.  When 1kHz stream delivery vary in time by 20 milliseconds (50Hz) the output of D/A converter will produce 1000Hz, 950Hz and 1050Hz - three frequencies instead of one (and many more at lower levels).  Amplitude of these additional signals will depend on range of time vibration, but it is very low.  It is still very audible, since not harmonically related to root frequency.  Many noise frequencies and many offended frequencies (music) results in added noise - less audible for random jitter (uncorrelated) than jitter induced by particular frequency (correlated).
Hey @mihalis I had the same MSB DAC, but the lowest model, the Discrete with the same ProISL / ProUSB module combo like you. At that time, there was an audible difference to my ears when feeding that DAC via USB from a NUC, then Nucleus, vs. a top server ‘ streamer I had then, the Innuos Statement. That’s what I heard, with my ears, my equipment, my room. That’s all I needed to know, and be happy, enjoying my music. I did not have to worry why, although I intuitively can figure out the reason, based on what the Innuos Statement has / does.

All what matters is we experiment for ourselves. Sometimes new things work, sometimes they don’t. We then move one to something that works, or just stay with what we have / own. It’s pretty simple. 
Also reading the earlier confusion about fiber optic. My bad, I didn’t want to detail the system and I should have had. I receive streaming via fiber and also play audio from local drives. All this is irrelevant to my point. My reference to fiber is only as it relates to what happens between the Roon core and the dac. My Roon core is home made and via usb it goes into something called usb pro which is basically msb’s conversion from usb to fiber optic cable. This cable then feeds the dac directly. I was only referring to this connection and not to the ISP. My understanding is that usb pro is clocked to the dac’s clock. And my search has been to figure out if I can improve on my home made nuc with expensive or well reviewed streamers. Stereophile just had such a review where the writer who seems quite honest clearly says that he can’t ascertain any audible differences. 
Hi kijanki, thank you for the clarification which I understand. 
My statement was a straw man assumption which I put forward for us to debate. It is “absolute” on purpose, like a math equation that we are trying to prove wrong. Sorry if I wasn’t clear about that and it sounded like I was saying that I know it is correct. Which I don’t.
The revised statement which is now informed by further feedback from two designer experts is that so long as we receive bit perfect signal via fiber optic cable through a converter that is clocked to the dac we get a signal that can’t be improved with expensive streamers. One said “everything before the conversion doesn’t matter at all”.
Some other arguments against this include that one may be introducing electrical noise to the house’s electrical system before the fiber and also that one may be introducing airborne ref. I suspect these are negligible in real life conditions especially as one can move the server to another room.

I repeat this is an assumption for debate although we now have many facts in its favor. And I hope it helps people save money and focus their spending on equipment that matters. 
What I am baffled by is 3-4 responses which were not only rude but also paradoxical: bashing an effort to emulate the scientific method whilst using such science based equipment. Maybe it’s the snake oil they put on their salad. 
All the best and thanks to those who tried to explain and help. M. 
@mihalis I called your post a "blanket statement" because you stated that nothing makes a difference when your DAC is connected thru fiberoptics etc.   There is a big difference between saying "I cannot hear the difference" and "Nobody can hear the difference".  There are many factors involved, including receiving hardware and the fact that your DAC is connected by fiberoptic is not changing anything.  I can only agree that with exactly the same data received in exactly the same format in exactly the same hardware to the same DAC - sound should be the same between providers.
Btw the designer of maybe the best steamer finally confirmed my straw man assumption. Turns out indeed it was correct. The signal is identical. No one can hear any difference.Of course I wouldn’t question that the person who was very offensive for no obvious reason earlier would hear differences from these identical signals from up there at his high horse. M.

Post removed 
Post removed 
Pased on the elegant prose in your opening statement, I have no idea.  It's not as you sound like a rambling nut case or anything.
HI yyzsantabarbara

Both Qobuz and Tidal use the Rest API for streaming. The following is a generic example of a streaming session using Qobuz/Tidal; Audirvana and a streamer/DAC using UPnP/DLNA.
Audirvana will issue a REST request over HTTP to the Qobuz/Tidal server. Your computer will create an TCP/IP connection to the Qobuz/Tidal server and transfer the request after the connection is established. Your computer will also use Ethernet to transfer the HTTP, TCP and IP protocol data units to your router for transmission over the internet. The Qobuz/Tidal server will respond to the REST/HTTP request over the TCP/IP connection. Once the response is complete the TCP/IP connection will be closed.
TCP/IP will provide error free sequenced packet delivery between Audirvana and the Qobuz/Tidal server. If any packets are lost it will automatically re-transmit them. Ethernet will provide error free frame delivery between your computer and a switch/router.
Audirvana uses the UPnP protocol suite to transfer the media content received from Qobuz/Tidal server to your streamer/DAC. It will use HTTP, TCP/IP, UDP,  Ethernet and other protocols.

You will get error free delivery from Qobuz/Tidal to your streamer. There can be reduced band width on your connection due to network congestion/errors.
@millercarbon
One of my good friends recently became an audiophile and has finally agreed that the most important variable is actually what you hear not what science tries to prove. 
One designer of streamers confirmed that his effort is to reduce any emissions via air, cable and power. My “strawman” above addressed cable. I don’t know about air especially since fiber allows us to move the streamers or NUC as much as a kilometer away! About power however I get the point. That could have an effect. The way I may have indirectly dealt with it is that I run the custom nuc with a linear power supply (150$ from China!) and I always isolate my front end (other than the nuc) with isoclean or balanced transformers. M. 
Hi Welcher. My understanding is that Ethernet sends in packets but since it is still an electrical signal it is possible it would affect sound although a well designed interface is unlikely to. Usb connections are proven to make a difference good or bad with measurements.