The Trouble With Audio Tests

“If your experiment needs a statistician, you need a better experiment.” ― Ernest Rutherford

I have not seen the results of yesterday's Amazing Randi/Ars Technica $1,000,000 challenge comparing an AudioQuest Vodka Ethernet cable to "a cheap Amazon Basics Cat5 Ethernet" but I can't wait to read the results and the analysis of those results. Should be entertaining. My guess is we'll be told that Ethernet cables do not make any difference even though many people have experienced otherwise. One real question is—when looking at results from audio tests involving more than one person, does it make sense to draw generalized conclusions that discount individual results?

Back in May, someone asked me to give these audio samples a listen to see if I had any preference for files upsampled using a linear phase filter or a minimum phase filter. I did one run-through and easily picked my preference for the first two samples while I found the third more difficult. It turns out I picked the minimum phase filter for 1 and 2 and the linear phase for #3. Again, with the first two samples the difference was easily heard and my preference was clear. So much so that I picked out the minimum phase filter in the second sample without having to hear the linear phase filter version.

Here's Archimago who offered the test:

"Remember that the technical discussion of digital filters can get complicated, often with no actual data around audibility to support positions. Over the years, special player software such has HQPlayer have been released with upsampling algorithm tweaking as a major feature. However, I hope that in using the open-source SoX algorithm with basic parameters that are not specially selected to reduce the 'ringing', this test can provide some insight into audibility in the 'real-world'"
For me, the audibility was clear cut yet here's Archimago's analysis of the results:
"As we know last week, statistically, indeed there was a preference for minimum phase filtering, but not strongly significant based on statistical data.

"I do believe it is worth remembering this when we read subjective reviews and comments in forum posts for example. Certainly, as it pertains to subjective preference for digital filters (even with as much ringing as the test here), it does seem that differences are likely subtle at best for most audio lovers even with what I believe is significantly above average audio equipment used in this cohort of 45 responses."

Well that all depends on who you ask. If you ask his data analysis you'll get a wishy washy answer. If you ask individuals who took part in the test, you'll get different answers. Take your pick to support your point.

Clearly any analysis that chooses to discount individual results in favor of the group result is to ignore the most basic and most important ingredient of listening to music on the hi-fi—our preference.

COMMENTS
johnnya's picture

If the sine non quon of audio testing is the individual response, then there is no point in entering discussion with those committed to statistical analysis of group differences (using an 'experimental' design) in choosing a preference via 'experimental" trials: the sole criteria lies in individual selection. There is no way to make problematic individual preference, no matter how much one might rail against a lack of scientific rigor; because,in fact, such rigor lies outside the domain of personal preference. It is a fools errand to argue otherwise. No one will change their mind, regardless of the 'evidence' adduced or the apparent rationality of their argument. The sole question is whether one determines to both trust their own hearing and whether they have the desire and financial means to make a purchase. Why audiophiles are stuck in the muck of their own persuasion and can't decentrate from their mind set, well that is question better raised in psychoanalysis. Whatever Randi demonstrates or fails to demonstrate is a friggin' waste of time.

VK's picture

!

jdaddabbo's picture

Agree.

otaku's picture

I would hope that Randi's test says either that there is an audible difference or there is not. A result that needs statistics would be a big disappointment. If it DOES show a difference, that is the end of the discussion (well, I suppose discussion would go on about cost and why there is a difference, but at least it would be rational). If it does not show a difference, I think that would still be inconclusive because it could be equipment dependent.

drblank's picture

always flawed. The room acoustics, ambient room noise, equipment, content, volume levels, listener's listening abilities, etc. etc. are all contributing factors on whether one can hear or not hear any differences between products. In addition, if there are additional cables/switchboxes that can alter the signal are present to minimize any differences between two or more products will also impact the listening test.

I prefer to just give myself enough time to listen to enough content to make up my own mind if I prefer one cable or product over another. Only in some occasions will I hear a big enough difference within a short period of time as it has happened, but I prefer giving a product enough time to fully audition the product.

I highly doubt Randi is going to have skilled listeners. What he should do is mosey on down to a mastering studio like Gateway Mastering and give Bob Ludwig some time to evaluate the product. At least he has a soundproof studio with good room acoustics, he has stelar equipment, access to a variety of master recordings and he definitely understands what to listen for. When it comes to audio equipment, your average Joe Blow simply doesn't have the skill set to do any sort of serious listening tests since they really don't know what to listen for. Now, what's the premise of the test in the first place? is it to verify that the average Joe Blow can hear a difference, a highly trained and skilled listener like a top rated and experienced mastering engineer or a specific person?

Clever Dean's picture

And those results track to some # out of X have a preference. That's my take away.

When he says "it does seem that differences are likely subtle at best for most audio lovers"

I don't read that to preclude other listeners that may have a clearer cut preference.

So his testing worked. I'm confused as to the over all point trying to be made here with the testing at Archimago's blog.

Michael Lavorgna's picture
...the only way to know if you have a preference is to listen for yourself. Saying that "it does seem that differences are likely subtle at best for most audio lovers" tells us nothing of real value.
jdaddabbo's picture

Unfortunate, but true.

Michael K's picture

There are relevant grounds for preference and irrelevant grounds. What the people calling for objective testing are really trying to prove is that some audiophile equipment preferences are based on something other than differences in sound of the equipment itself. I guess they suspect there are criteria at play that are irrelevant to the sound--like price (i.e. status via conspicuous consumption) the opinion of others people with agendas beyond just the sound of the equipment (sales people who may stand to make a higher commission or bonus for pushing one product over another, or journalists whose copy will be controversial and bring in more subscribers, etc.). They want to save us from ourselves and show how we are manipulated, the pompous jerks. Anyone who has passed Psych 101 knows that those other, irrelevant grounds for preferences can probably change the perceptions of at least some people for some amount of time (statistics). Just like relevant grounds for preference, irrelevant grounds can be very convincing and even exciting for awhile. But, as I have occasionally have seen in myself and others, with additional time listening to new equipment purchased for irrelevant reasons, the grounds for preference fade away. And we start looking for the next new/best thing that can bring back that experience of the manipulated preference. Sounds a bit like audiophile behavior...or in an extreme case, addictive behavior. There is, I would argue, one objective test that might be able to settle the question of relevance of the grounds for a preference: a comparison of the wave form of the same music sample through two systems that are identical except for the component under test. Then subtract one wave form from the other to see if there is a physical difference that is causing the perception. The software and hard ware to do this is available to the serious hobbyist. It won't say anything about what two people perceive or prefer, only if there is a physical reason in the output of the system that could be the grounds for the perceptions and preferences. If there is, then that's fine, if not then there is some else that is affecting the psychology of the people and so affecting their perceptions and preferences.

Michael Lavorgna's picture
Testing for preference is a fool's errand, imo, since the very nature of a test, double-blind or A/B/X, is unnatural. It is not the way we listen. Here's neuroscientist Daniel J. Levintin from This Is Your Brain On Music:
The scientific method requires that we control all possible variations in order to be able to draw firm conclusions about the phenomenon under study. Yet such control often creates stimuli or conditions that would never be encountered in the real world, situations that are so far removed from the real world as not even to be valid. The British philosopher Alan Watts, author of The Wisdom of Insecurity, put it this way: "If you want to study a river, you don’t take out a bucketful of water and stare at it on the shore."
When testing for difference, the idea being that the only explanation for hearing a difference between, say, two cables is placebo, we cannot erase even one positive result with statistical analysis. If just one person hears and correctly identifies A from B under test conditions, this proves that there is a difference.

So while I understand the evangelical zeal of some self-described objectivists, there is no substitue for listening over time to determine what we prefer listening to.

In terms of measurements, I know of no one more experienced than John Atkinson of Stereophile in performing measurements of audio gear while also providing subjective listening evaluations. Here's a favorite quote from John:

"But, I am sure, to those engineers who developed such tests 50 years or more ago, it was still the whole experience that mattered, the measurement being used merely to aid diagnosis than to determine the quality of the experience. Now, however, in the latter years of the 20th century, a whole culture has arisen that insists that the measurements are the experience. If ever I came across a worse case of mistaking the messenger for the message, I have forgotten what it was."
Bromo33333's picture

If you cannot have a clear unambiguous objective measurement free of preference, through a measurement of some kind, it's nothing more than a vague marketing study. Nothing more, nothing less. If you have nothing to market, then you have wasted your time as well as your subject's time.

My issue with the ABX test is that in the end, the results are a tabulated survey where you ask the people participating what they heard and liked. That is a weak test, and to give it more weight than a pile of opinions (since you are trying to find out if A or B is objectively superior in some way).

If you could measure the brain response and notice a difference between A and B, then you have a good ABX test. Otherwise? It isn't engineering, and if a science, it's one of the soft ones.

absentidei's picture

"we cannot erase even one positive result with statistical analysis. If just one person hears and correctly identifies A from B under test conditions, this proves that there is a difference."

Eh, only if that person can consistently hear the difference in a large number of tries, without knowing what cable they're listening to.
Our confirmation bias is HUGE.

So if you KNOW what cable you're listening to, you'll hear a difference.

Do you think you'd be able to identify the more expensive cable at a higher rate than guessing in a proper double blinded listening test?

Michael Lavorgna's picture
Not under test conditions with a significantly correct number of responses. If one person passes the test, done dealio.

Here's what I think about blind tests - they are a waste of time. Ars has done me the favor of proving my point.

absentidei's picture

"If one person passes the test, done dealio."

And so far, as far as I know, no one has been able to do that when it comes to ethernet cables...

Michael Lavorgna's picture
...a proper test. So to say "And so far, as far as I know, no one has been able to do that when it comes to Ethernet cables..." is making a pointless point.

The Ars/Randi "test" fiasco, in all of its layers of incompetence, illustrates that these kinds of tests are not easy to pull off if you want valid results. Frankly, the notion that my saying the differences I heard between different Ethernet cables three years ago were easy to hear has caused sites like Ars to spend so much time and money to debunk strikes me as being rather pathetic.

absentidei's picture

Well...
Just the fact that the people who sell these claim that they are directional show that they are lying to their customers...

If you could easily hear differences between them, why can't anyone else do that in a blind test?

And frankly, I don't believe you could do it in a double blinded abx test either.

Michael Lavorgna's picture
...should have been set up to debunk directionality.

No one has ever attempted to do this in a blind test as far as I know. If you are aware of such a test, please share the link.

"If you could easily hear differences between them, why can't anyone else do that in a blind test?"
Here, let me repeat myself, "That's because there has never been a proper test. So to say 'And so far, as far as I know, no one has been able to do that when it comes to Ethernet cables...' is making a pointless point."

Context. It's a very important concept in the world of hi-fi. My results were based on me listening to my music in my room with my system. This is why I never say something "You will hear...". Which is also why I recommend listening for yourself in your system. Ideally over time. Listening to one 30-second snippet 3 times on an unfamiliar system, as in the Ars/Randi "test", tells you exactly nothing of value. Especially when the test system hiccups and injects a 30-second delay between 30-second music snippets.

Double-blind test are a waste of time, imo, and offer zero relevance to the listening experience. So you'll never know if I would pass one because I consider them worthless.

VK's picture

If the only thing that matters is our own experience, and group tests are wrong, sites like Audiostream and many others are tottally futile. But they are not futile. So why try so hard to prove your point? We all get it, you have a strong opinion about how someone must feel pleasure in a audio system. Personally i would like to see more audio reviews and tech articles of how to take the best of our gear instead of having to see these "preach my point of view" articles.

As always, best regards!

Michael Lavorgna's picture
We share experiences.
We all get it, you have a strong opinion about how someone must feel pleasure in a audio system.
I've never said that. The purpose of hi-fi is to enjoy music.
VK's picture

...fair enough!

"The purpose of hi-fi is to enjoy music."

I agree. But our modern hi-fi is the result of many tests with plenty of professionals and "golden ears" involved in the process (among other things, of course).

Best regards and a nice weekend!

Michael Lavorgna's picture
That hopefully will more clearly describe my approach to reviewing in greater detail. But I'll make sure not to spend too much doing so ;-)

Enjoy your weekend.

audiofool's picture

loving audio does not mean you have the training to make good listening judgements. My early days of buying audio gear were lemon filled from lack of listening experience to know what to audition for

ednaz's picture

I can hear things in audio equipment, and in music, that others can't. I spent years studying music and then trained and performed as an actor, which again puts a very high value on auditory discrimination. I also have unusual high frequency hearing; at my age I can still hear CRTs, power conditioners, and other high frequency noise. I can tell the differences that some cables make in my systems, that changes in vibration isolation have made, and differences in bit depth and sampling. Can hear a lot of things in some systems, but in others, not so much, because some of my systems are much more sensitive and resolving than others, and in different ways.

I'm perfectly OK that a lot of people can not tell many of the differences, and impressed when something breaks through. Upgraded my DAC recently, and my wife, who professes that she can't tell much difference from any component change (and since most of what she listens to is spoken word through the cheapest earbuds she can find... I tend to believe that.) The new DAC totally blew her away. So far, even my least hearing sensitive friends have noticed.

But I'm OK if I can notice things they can't, because as much as I love wine, and better wine, for me a cabernet is a cabernet. I have good friends whose enjoyment of wine is based as much on their ability to not only identify the grape, but the year, region, and vineyard. I have no doubt they taste things differently and more deeply than I do. Doesn't make me enjoy wine less, and when I tried to train myself to taste more differences, I wasn't so good at it.

I think the audiophile community feeds the beast by insisting that if those who say they can't hear a difference would just try harder, they would hear a difference. Maybe not. And why care? Their enjoyment (or lack thereof) doesn't affect yours. I tend to smile and shrug and feel for them the way my wine friends feel for me. Way too many people insisting that other people are idiots if they don't share our very subjective abilities and personal opinions.

Reed's picture

I think if most people would take the money they spend on cables and apply it to room acoustics they would get about 1K times more bang for the buck, achieving some night and day improvements vs. "listen to the same track 57 times...I think I hear a difference" kind of deal.

Vigna ILaria's picture

The trouble with tests like the upsampling one you described is not so much the validity of the test itself, but the validity of the conclusions drawn. In this example you compared two filters, one with linear phase and one with minimum phase, with mixed results. But if you had found a unanimous preference for, say, the minimum phase filter, you would probably consider this a plausible demonstration of the superiority of minimum phase filters.

In reality, all you have proved is the superiority of Filter A over Filter B. It is a more knotty challenge to demonstrate that the "Phase-ness" of the filters was responsible for what you heard. This is something that interests me.

dpudvay's picture

You said, "when looking at results from audio tests involving more than one person, does it make sense to draw generalized conclusions that discount individual results?"

Due to the subjective nature of the tests, the answer would be yes if the individual results are not obtained in a method that removes inherent biases of the observers.

A methodology that allows the observer to know the cables under test does not remove those biases. You claim ABX and DB testing is not how music is listened to in the real world. Surely a test could be designed where one person set ups your system such that you don't see electronics, only the speakers are place visible to you. This is especially easy in the case of Ethernet cables. You can listen for as long as want and write your review without ever knowing what it was you were reviewing. And then you can do so again with another set of cables. This isn't changing the way you listen to music, but it does remove confirmation bias from the tests. Since the judgement of the performance of cable is purely auditory, seeing the amplifiers/sources/cables etc., is unnecessary for real world listening tests and hiding them from the reviewer would bring credibility to your testing.

Michael Lavorgna's picture
...requires preconceptions. Back when I reviewed the AQ Ethernet cables I had none. As a matter of fact, I was very skeptical that an Ethernet could have any effect on music playback. I had not read any others reviews, as far as I know there weren't any at the time, and the only communication I had with AQ was the improvement they heard was "sick".
cundare's picture

Earlier this year, I sent Mike Fremer & Art Dudley a copy of a 1980s MIT Computer Music Journal article in which the researchers performed comparisons of speaker cables. The most interesting thing (to me, at least), is that statistical analyses of the results showed virtually not difference in confidences between a single-blind and a double-blind test. (It also demonstrated that there was a clear audible difference between cables. One listener, identified only as "JA," actually scored 100%.)

I wonder -- has anybody ever conducted a double-blind test to determine whether double-blind tests are more accurate than any other type of testing?

Michael Lavorgna's picture
...the people most adamant about testing already know the results.
cundare's picture

Seriously?

cundare's picture

If you're trying to look smarter, better educated, or more credible than you really are by conspicuously inserting Latin into a forum posting, a sine qua non of that effort is to at least come close to quoting the Latin correctly. Unfortunately, I went to Catholic High School, so a disproportionate number of my brain cells are wasted storing Latin grammatical rules & vocabulary.

Music_Man's picture

I’ve spent enough time reading “Archimago's Musings” to recognize that any additional time spent, would be an utter waste!

I would rather read subjective posts from folks whose opinions I DON’T trust, than to any more so called objective conclusions from “audiophiledom’s” Mr. Magoo….!!

I frequent audio forums for subjective opinions, not the opposite. I’ve found that I can, with enough samples, begin to formulate what and whose opinions I favor and as such, find audio equipment directions to delve deeper into. In fact, the 10’s of 1,000’s of subject opinions and conclusions I’ve found on forums have enabled me to build a system that would have been impossible otherwise!

So the last thing I want to be is saved from myself by audio's self-appointed apostles of objectivity and placebo awareness. I know it may be hard for them to believe, but I understand all that the objectivist’s seem to assume and believe that I don’t; and yet, I never allow measurements, DBT, a component’s specific part’s or its overall design philosophy, to rule the day. I use only my ears in the buying decision mix and as such, I buy what my ears and ultimately my emotions prefer; which unsurprisingly, is how I also select the musical types and performances that I buy and play!

Maury's picture

The basic problem with audio tests is that we have no agreed on standard recording and standard playback system that serves as the marker of all deviations. Each "audio truth detector" comes up with their own recording and says what differences are there playing it through A & B? Well, different recordings will be affected differently by different system changes and that says nothing about differences between listeners.

Let's step back and think of normal audio experience. When we make changes to a familiar audio system playing a set of familiar recordings we typically hear the effect of those changes more on some recordings than others. Over the years I have made all kinds of changes to my several systems. Some familiar recordings have been transformed either positively or negatively. Others sound fairly similar through the changes, without being identical of course. And yet objectivists would say all the equipment from version 1 to version 20 was vastly more capable and accurate than it needed to be to play the music without significant distortion.

Nor does this require golden ears. Extremely non audiophile listeners can easily hear most differences, which I know from personal experience. But it depends on the kind of material and the nature of the change whether they will immediately hear the difference. This is the problem with the testers. They are amateurishly scientific. They try to skip the 20 preparatory steps needed to properly answer the question that they want to know. They measure what is convenient to measure, not what they need to measure.

rajivSK's picture

I have some great audio gear, spent years of my life, listening, playing, producing.. I get it, I really do.

The problem is, I'm also a software engineer. I know what happens to audio in a digital environment. The 1's and 0's being sent through your ethernet cable either get to the end in order and reproduce the original audio buffer or they don't. If they don't arrive as they are meant to, they do NOT deteriorate your audio, they create gibberish. Complete and utter gibberish that sounds like someone bumped into a record player. Lucky enough for us, any old ethernet cable is more than capable of handling audio streams, even at the highest bitrates, without any problem.

Oh by the way, this jitter and popping is only possible when your audio is being sent as raw bitstreams, which is very, very unlikely. It is almost certainly being transferred through the TCP protocol, which checks the validity of all the received packets before further processing. If a packet fails, a new one is requested until it is received as an identical set of ones and zeroes. If the packet is not able to arrive on time, due to multiple failures, caused by interference for example. The audio doesn't deteriorate either. Instead it just stops playing at all.

So unless you are experiencing violent popping and jitter, or complete failure to play audio at all, your ethernet cable is transmitting a perfect, untouched, exact to the original at the single bit level, digital audio stream.

These people are trying to scam you out of money that could be spent wisely on some actually valuable audio gear.

Michael Lavorgna's picture
...the integrity of data transmission.
rajivSK's picture

Exactly! It either is, or it isn't. It can not be better or worse. This cable doesn't even require any A-B testing.

Michael Lavorgna's picture
...bit disagree on your conclusion.

Here are some thoughts on why: Digital Cables and Noise

rajivSK's picture

The question is not if digital cables can be susceptible to noise, they can, without a doubt. The article you posted talks about trying to minimise that noise, which in itself is very important for carrying signals with high confidence. It however isn't possible to send any signal completely noise free. That is why ethernet packets are verified upon arriving at the destination. You could compare it to a very intricate lock and key system. If the arriving data has been affected by noise, the key won't fit the lock anymore. At this point a new packet is requested.

Keeping noise down is important because requesting packets over and over again introduces the need for larger buffers, which increases latency. Only once a packet arrives that has been verified will it be put into the audio buffer to be played. If it passes the check it's perfect. Any noise introduced at cable level only increases latency, it does not deteriorate the signal.

rajivSK's picture

I feel like I need to add a couple of details for clarification.

An Ethernet(or USB) cable is just a cable, like all other cables, it is susceptible to noise.
Digital signals are just signals, like any other signal, they are susceptible to noise.
The TCP protocol has verification in place to discard any packet that is affected by noise.

USB interfaces, headsets, or any other peripheral can use any kind of proprietary protocol for transmitting digital information, which may or may not have a verification system in place.

Audible noise can occur with some of these, lesser engineered, peripherals.

Your NAS uses the TCP protocol, not a proprietary protocol. This is why we can say beyond any shadow of a doubt, that the cable talked about in this very specific instance, will not benefit your audio signal in any way.

Michael Lavorgna's picture
What I am suggesting is that noise, separate and apart from the data, can in fact travel on a digital cable and enter the attached device. I give a number of examples of this in the article I linked to.

I am not questioning the ability to pass data. That would be foolish, imo. What I am saying is that cable-born noise can enter an attached device. With a network-attached hi-fi, which is what I'm talking about, this noise can affect the resultant analog signal. This is not a fact by any means, nor do I believe it is the only thing going on, but it is a starting point for understanding how different digital cables can effect the sound coming out of connected analog devices.

You raise an important point:

Keeping noise down is important because requesting packets over and over again introduces the need for larger buffers, which increases latency.
Some people have suggested that the more processing the music server, a computer in this case, has to do, the more internal noise will be generated resulting in poorer audio performance.

The fact of the matter is, I can not tell you exactly why I heard a difference between various Ethernet cables. What I can say is over a period of months of listening in my room and in my system which I know very well, I did in fact hear a difference. You are of course free to make of that what you will.

My recommendation for people with a network-attached hi-fi is to buy good quality CAT6 or CAT7 Ethernet cables and try them. If you find no improvement, return them.

rajivSK's picture

Ah, my mistake about the noise entering your system. Although minimal, I can see your point.

The CAT rating however, is not a factor. Neither are any of the other claims made by the manufacturer because the noise may be carried in many ways, including on the shield itself. These are my main issue and give me a bad taste in the mouth.

If you are experiencing noise from down the line, the best way is to isolate the electrics completely and run a fiber optic cable to your player. Still cheaper than this cable and will definitely eliminate any noise other than that from it's own power supply.

Michael Lavorgna's picture
At the time of the first review, I did not see how an Ethernet cable could make any difference. I worked in IT for nearly 20 years and the company I owned installed and certified Ethernet networks. This is not to say that I am an expert by any means. This way many years ago and I was not directly involved in the actual testing etc. I did however talk to my friend who ran this business for us and can retell countless stories of botched Ethernet installations, many due to poor terminations and poor quality cables.

I own two short lengths of the AQ Vodka Ethernet cables and have been using them for for years. I would not use fiber for a 1m run.

"The CAT rating however, is not a factor. Neither are any of the other claims made by the manufacturer because the noise may be carried in many ways, including on the shield itself. These are my main issue and give me a bad taste in the mouth."
How is the CAT rating not a factor when we're talking about, among other things, improved noise rejection? Also, what specific claims have been made by AudioQuest that you find objectionable?
rajivSK's picture

"How is the CAT rating not a factor when we're talking about, among other things, improved noise rejection?"

I thought we passed the noise on the signal part, you were talking about electrical noise directly influencing your analog signal at the player. The only plausible type of noise I might add.

This type of noise is not a concern for ethernet cables and for this the CAT rating is not intended. Noise carried on the shield itself can just as well influence your player, the heavier the shield, the more likely it is to pick up on interference and carry it to your player.

The length of the run, when talking fiber optics, is irrelevant. The fibre optics create a separation between the two electrical systems and therefor can be beneficial even for a run of only 1cm (highly impractical of course) just to separate them electrically. Way more beneficial than any type of ethernet cable can ever be.

Michael Lavorgna's picture
...simply cost more when I looked into it years ago and the AQ Ethernet cables worked for me.

As far as your other comments, I'm not following you. Whether or not a cable's shield will carry noise "into a player" is dependent on a number of things including the cable's grounding scheme. The tighter tolerances in twisting ratios in higher CAT levels equals less internal "noise". Where my disconnect comes with what I think you're saying is I don't understand why you say that signal-based noise is not an issue in Ethernet when it is.

When you say "The only plausible type of noise I might add." I'm assuming you're referring to electrical noise so sure. Who said otherwise.

I feel like I'm falling down the rabbit hole ;-) I'm not sure where this conversation is going but do me a favor and cut to the chase. I'm looking forward to a beer and some dinner and a relaxing weekend.

JVan23's picture

Individual preference can be swayed by inaccurate perception. Inaccurate perception is a documented phenomena. Feeling good about gold plated printer cable connectors does not result in more vibrant pictures on the page.

At the end of the day practicality should come into play. If you are told that the cable cannot make a difference then it is best to put aside the perception of better sound and spend your time and money on something that will make a difference.

In the original article, time was mentioned. Also, the manufacturer seems to imply that the cable being more slippery for the electrons makes the music flow faster/better (paraphrasing). This is misleading.

At the end of the day the data has to get from A to B and be ready to play. This is not a just-in-time system, there is buffering involved, verification of the data, decoding of the digital data, all occurring well before any sound comes out. Depending on your media player it could download the entire song from the NAS before the first note is even played.

The only possible point of merit is noise adding to the system. But there are two problems with that. One, the Ethernet connection is likely nowhere near the DAC, both in terms of the sound pipeline and physical distance, and two, the computer itself has countless avenues for noise introduction. Reducing one source of noise is pointless.

Michael Lavorgna's picture
...with internal DACs that are connected directly to your NAS/switch via Ethernet. The same goes for some music servers.

In the case of my setup for these Ethernet listening sessions, I was using a MacBook Pro as my music server running media server software that does in fact load everything into memory prior to playback. The Ethernet cables in use system-wide included one from the router to the switch, one from the NAS to the switch, and one from the switch to the MacBook. A USB cable connected the MacBook to the external DAC.

To be frank, I'm really not interested or knowledgeable enough to discuss how noise travels through this entire system.

The point of contention is this: I heard a difference between different Ethernet cables in this system. If you want to discuss this, I'm game but I've doing so for nearly three years so my patience is admittedly thin.

JVan23's picture

Ethernet packets are transmitted with multiple layers. The receiver hardware takes care of some of the decoding, rearranging into the correct order, requesting retransfers in case of errors, etc. At some point you get into driver software running on a CPU that reassembles the data packets into the original file before it is transmitted to the DAC.

There is no possible way for the DAC integrated circuit chip to directly connect to the ethernet cable. My guess is you are referring to the overall DAC device which is still a computer that performs all of the above processing and assembles the entire file or a good portion before it is sent off to the hardware that converts the digital stream to analog, amplifies, etc. This is the reason why timing on the Ethernet cannot be a factor.

Either you have a high end device for the digital to analog conversion and amplification or a generic device (e.g. PC). Either way they both get the exact same set of 1's and 0's for conversion (assuming the player software isn't adding any WOW enhancements). In the case of a lower end device the quality of the amplifiers and digital to analog circuits would negate any possible noise improvement from the distant ethernet connection. In the case of a high end device there should be sufficient isolation to cancel any effects of that same connection.

I think the actual point of contention is whether you heard a difference or whether you thought you heard a difference.

If there is a difference there should be no need for arduous tests involving connecting and disconnecting the cabling. Just play the song directly from the storage of the DAC device vs streamed from a NAS to hear the "difference".

Michael Lavorgna's picture
...with music stored on a NAS, which is what we're talking about; use a computer with an external DAC, use a network player with an internal or external DAC, or use a music server again with an internal or external DAC. Of course an Ethernet cable does not connect to "a DAC", no one said they did, but in the cases where the player/server has an internal DAC, the Ethernet cable is in fact connected directly to the device performing D to A conversion. There is a DAC, the Merging Technologies Nadac, that accepts Ethernet as an input but this is obviously an exception (it uses the Ravenna protocol).

Timing is not an issue, imo, for a number of reasons including the fact that in most cases the data is stored in memory prior to playback. I have repeatedly said I am not talking about problems with data transfer. Regarding noise in a mixed signal system, there is no way to adequately explain away this possibility in a paragraph. Thankfully, we have a very knowledgeable reader on this subject and he's shared many informative posts. I'll point to "CG" comments here on AudioStream as a good starting point in better understanding the issues involved.

"I think the actual point of contention is whether you heard a difference or whether you thought you heard a difference."
I agree - this is the point of contention. The logical real-world outcome of this absurdly contentious contention is that people either will think it's worth their time and money to try CAT6 or CAT7 Ethernet cables or not after reading my reviews. That is what we're talking about. Period. The rest of this silly "debate" is puffery and clickbait "journalism".