[SoundAndMotion]

@MrHaelscheir started a thread with a great first post that resembles a mini scientific paper, with intro, methods, results and discussion. An important component is his inclusion of his own measurements. If interested in the topic, I recommend reading the first 2.5-ish pages of his thread.

As those familiar with forum discussions know, thread drift often occurs. Sometimes it drifts to nonsense, but sometimes it drifts to a new, also-interesting topic, as happened with that thread (drifted to various issues surrounding (lol) spatial audio). @MrHaelscheir noticed this, but sub-forum moderator @castleofargh pointed out he can't create a new thread with the relevant new posts moved to it.

Because I don't like threads with multiple interwoven topics, I started this to continue @mrHaelschire 's thread discussing @WaveTheory 's video series: Cables Matter: The Science Behind Hearing Differences in Audio Cables. The series interested me because it is unusual for one who hears differences in cables to have his strong scientific/technical background (professor/scientist with dual PhD's in Physics and STEM Education). Only Milind Kunchur comes to mind as another.

I watched all the videos, and comment in the next post.

 

[SoundAndMotion]

@WaveTheory 's video series is called: Cables Matter: The Science Behind Hearing Differences in Audio Cables.
There are currently 4 videos discussing the topics and 2 with multiple mini reviews of cables. He plans 2 more videos discussing the topic.

Whereas @MrHaelscheir and others praise WaveTheory's treatment of science, measurement, waves and cable electrical properties, they are less comfortable with his ideas about pattern recognition. My impression is reversed from that.

Pattern Recognition (episode 4 of the video series)
Both main branches of pattern recognition (PR) research (artificial neural networks and real brains, aka cognitive psych/neuroscience) are over 50 years old and the concepts have been applied to virtually every sensory modality, including hearing. One of the reasons PR is interesting to me, is that PR often evades the reductionist approach to science. This is because many systems within the brain, including sensory perception, are nonlinear, which often leads to the whole being greater or less (!!) than the sum of the parts. Not only are the systems nonlinear, but the exact nature of the nonlinearity depends on "attention", which can make a controlled experiment quite a challenge! Therefore, many ideas that it's not possible to hear that can be incorrect. The cocktail party effect is a great example.

WaveTheory correctly points out (but may shoot himself in the foot) that there are several possible outcomes within PR. For example, 4 possible outcomes might be:
A- Pattern exists* and is detected (PR success)
B- Pattern missing* and not detected (success)
C- Pattern exists but not detected (failure)
D- Pattern missing but is "detected" (failure)
He believes the large number of people NOT hearing a difference in cables is an example of outcome C, but some people (such as himself) succeed in getting outcome A. It is certainly possible that this is true. (*When I say exists or missing, we need to consider whether we are capable of detecting the pattern. See below in my next post.)

So how can we tell the difference between A and D? WaveTheory puts forth the idea that we should determine whether the pattern exists or not. He proposes a mechanism for its existence in episode 3 about AC Signal Transmission (see my take in my next post).

WaveTheory plans 2 more videos: A summary of the 1st 4 videos, with the addition of a "testable evidence-based hypothesis", and a rebuttal of any rebuttals he gets. I'm very interested in the "testable evidence-based hypothesis". I have an idea what I would suggest, but I'll see what he says first (Ep. 5 coming soon IIRC).

 

[SoundAndMotion]

I was very interested in @WaveTheory 's take on this topic, mainly because many keywords from his background also apply to mine. Alas, I was a bit disappointed.

His pedagogic choices for presenting the material to the lay public are not the same as mine in Episodes 1 and 2, but you say toe-may-toe, I say toe-mah-toe (credit "Let's Call the Whole Thing Off"), and some useful stuff was explained. I'm not talking about the time vs. phase issue brought up in the original thread.

I have some problems with Episode 3. Rather than rant and belabor the issues, I'll just say:
Electromagnetic waves are different from an electrical signal representing some other quantity (e.g. sound waves).
He says 4 times in the video that the "Velocity Factor" graph shows "measurements", when it is a MATLAB graph of the idealized, theoretical equation. You can download the MATLAB script and graph it yourself.
He shows the following graph in this video and again in the Pattern Recognition video to show different patterns possibly produced by the cable:

I don't know anyone with a PhD (I know many), let alone a PhD in Physics (I know a few), who would copy and paste a graphic of the 2nd wave and shift it to represent the time shift he had previously calculated (= 5 nanoseconds), and then scribble a wave by hand showing the "changed" superposition, i.e. the new "pattern". It is trivial to produce the graph correctly. The top wave represents 2.5 cycles of a 1kHz wave and the 2nd one is 5 cycles of a 2kHz wave. So the whole graph is 2.5 milliseconds wide. FYI, there are 1 million nanoseconds (ns) in one millisecond (ms). He hand-shifted the graph about 1/45 of the width of the graph, so about 11000 times more than 5 ns. The scribble on the bottom is even worse.
Had he produced an accurate graph, and you viewed it full-width on an 8K screen, the 5 ns shift would be 1/65 of a pixel. On a more common Full HD screen, it would be 1/260 of a pixel. I don't think you could detect the difference in pattern, even on a giant screen with a magnifying glass. Also, an accurate graph would shift the 2 kHz wave left, not right.

@gregorio mentioned in the other thread that before discussing the audibility of phase distortion (a controversial topic), one must first verify that phase distortion exists. A 5 ns shift represents a 0.0036 deg phase shift of the 2 kHz wave, or a 0.0018 deg shift for the 1 kHz wave.

IMHO, WaveTheory has shown that there is not a detectable pattern change due to cables.

 

[VNandor]

I think that the suspiciously linear downward slope (up to like ~5kHz) of the HE-1's phase response just indicates a simple delay somewhere in the measurement loop. I wouldn't really consider that phase distortion.

I'm very sorry for saying this as well but I find this way of measuring phase distortion caused by cables very, very dumb. Instead of measuring the analog signal that goes into the headphone directly, maybe with a cheap soundcard that would have a tiny amount of (both linear and nonlinear) distortion and noise, the signal first gets converted by a headphone that has a significant amount of both linear and nonlinear distortion, gets coupled to a microphone, where both the microphone and the coupling have their own ways of further distorting and adding noise to the signal and all that for what? So the analog signal generated by the microphone could go into the very same soundcard input that could have been directly used all along except this time the mic level signal will probably have to go through an other amplification for even more distortion and noise. I can't think of any reason to do a measurement this way but feel free to bring them up if there's indeed any.

The reason I'm criticizing this part of the post instead of the other one is because I think commenting anything about these "theories" about why headphone cables sound different is largely pointless.

 

[SoundAndMotion]

I think that the suspiciously linear downward slope (up to like ~5kHz) of the HE-1's phase response just indicates a simple delay somewhere in the measurement loop. I wouldn't really consider that phase distortion.

I agree that below 5kHz it does look interestingly like simple delay. The problem with that idea is the phase wiggles a bit between about 7-9kHz and 13-15kHz but is mostly rising (a lot!) above 9kHz. Do you believe the simple delay knows to begin stopping the delay above 5kHz? A frequency-dependent time delay is... phase. So although this is very pathological phase distortion, it may include a simple time delay.

I'm very sorry for saying this as well but I find this way of measuring phase distortion caused by cables very, very dumb.

I understand this point, and you certainly make a powerful argument. But there is also an argument for testing the whole setup in-situ. Your method would not have shown an extremely important result: retesting the same cable produces similarly large changes to the result as the other cable does. So we see that moving you head or adjusting the headphone on your head washes out the difference in cables.

The reason I'm criticizing this part of the post instead of the other one is because I think commenting anything about these "theories" about why headphone cables sound different is largely pointless.

Point well taken, but when responding to WaveTheory's theory, one has the opportunity to help him understand it's weaknesses. So, pointless for you, but not everyone.

 

[VNandor]

I agree that below 5kHz it does look interestingly like simple delay. The problem with that idea is the phase wiggles a bit between about 7-9kHz and 13-15kHz but is mostly rising (a lot!) above 9kHz. Do you believe the simple delay knows to begin stopping the delay above 5kHz? A frequency-dependent time delay is... phase. So although this is very pathological phase distortion, it may include a simple time delay.

I'm suggesting that if you accounted for time delay, the HE-1 would be minimum-phase, or close to minimum-phase at lower frequencies just like a dynamic headphone would be. I have to admit I didn't look at how massively the phase shift rises at the higher frequencies. This certainly isn't typical as far as I know.

I understand this point, and you certainly make a powerful argument. But there is also an argument for testing the whole setup in-situ. Your method would not have shown an extremely important result: retesting the same cable produces similarly large changes to the result as the other cable does. So we see that moving you head or adjusting the headphone on your head washes out the difference in cables.

If the frequency response changes caused by the cable are supposed to be measured, then this is what should be measured and not something else. This method is unsuitable for that because the changes in frequency response caused by different cables are orders of magnitude lower compared to the deviation caused by this measurement method. If someone wants to know if the tiny change in frequency response is relevant, it's not unreasonable to try and see if a simple reseating causes a much bigger change than the one caused. My "problem" is that there's a significantly better way to measure changes caused by cables that's also much simpler and cheaper that gives a more accurate answer to the question, "do cables cause changes in FR". I suspect a direct measurement done well could be accurate enough to show that it indeed does cause a change in FR and it would also show by how much rather accurately. You could also get a definitive answer to "does it matter" because (I assume) the changes shown would fall way, way below of what's audible. I could do it myself in a weekend but I think it's a complete waste of time, I wouldn't learn anything from it, and I doubt anyone actually cares what some guy on the internet measured.

 

[gregorio]

Whereas @MrHaelscheir and others praise WaveTheory's treatment of science, measurement, waves and cable electrical properties, they are less comfortable with his ideas about pattern recognition.

In my case I certainly would not praise his treatment of science, measurement, waves and cable electrical properties. Although generally the basic explanations are accurate, he still pops in a few howlers here and there. For example, his definition of an amp: “Amplifier creates energy differences (voltages) that flip back and forth” - Not unless it’s broken it doesn’t, an amplifier should absolutely not create energy differences, it should only amplify the differences already created. (Really, a professor with a PhD in physics doesn’t know what an amplifier is/does?). While problematic in places, his basic explanations of waves, science, sound, etc., are a cause for relatively little concern compared to the relatively hugely troubling assertions he concludes from those explanations. Many/Most of which he makes in the Pattern Recognition video (part 4). For example, effectively: Different cables produce TINY differences (in patterns), Humans can detect TINY differences in patterns (in some cases better than any other animal), ergo humans can hear those TINY differences between cables. Of course, this is completely fallacious thinking because “tiny” is just a relative term not an absolute value, otherwise: The Earth is tiny compared to the sun, a baseball is tiny compared to the sun, ergo the Earth and a baseball are the same size and humans can play catch with planets! If physics/physicists were not aware of this obvious fallacy there pretty much wouldn’t be a science of physics. And, maybe it’s just coincidence but this fallacy (of scale) is one of the oldest of audiophile fallacies?

IMHO, WaveTheory has shown that there is not a detectable pattern change due to cables.

If you ignore the 130 years or so of scientific investigation into audible thresholds, which @WaveTheory has (on the basis that he looked and looked and couldn’t find any), then he has shown that there *could potentially* be a pattern change audible to humans. His only evidence that it is audible is his most unscientific sighted test which he performed on himself and his wife, leading to another of the oldest, most stereotypical of audiophile fallacies; “even my wife could hear it”. Just another coincidence? How many coincidences have to line up before the probability of it being coincidence is virtually zero? Not to mention, we of course can’t just ignore 130 years of science.

I don't know anyone with a PhD (I know many), let alone a PhD in Physics (I know a few), who would copy and paste a graphic of the 2nd wave and shift it to represent the time shift he had previously calculated (= 5 nanoseconds), and then scribble a wave by hand showing the "changed" superposition, i.e. the new "pattern".

I also have worked with and known many with PhDs and also don’t know of any who would have done that. Some undergrads probably would (but not many) and they certainly wouldn’t for a Masters, let alone PhD and in fact it would be the job of a professor to recognise and comment on students who did that! Of course, we can all make a mistake, even a PhD and professor, but not mistake after mistake, fallacy on top of fallacy, completely unscientific test methodology, unsupported assertions and conclusions, not to mention looking and looking and failing to find not just a single or couple of papers but an entire body of scientific research, how could you even pass a degree, let alone a masters and then a PhD, if you can’t find an entire body of research? Each of those is inconceivable but all of them together?! What is entirely conceivable (even highly probable), due to the series of “coincidences” of relying on stereotypical audiophile fallacies, is that he’s actually just an experienced audiophile who’s done a bit of research into the basics of sound waves, analogue signals and pattern recognition in order to justify his erroneous audiophile beliefs and does not have the qualifications/position he claims.

I suspect a direct measurement done well could be accurate enough to show that it indeed does cause a change in FR and it would also show by how much rather accurately. You could also get a definitive answer to "does it matter" because (I assume) the changes shown would fall way, way below of what's audible.

Such a “direct measurement“ has been done here on this forum. I believe Nick Charles measured a bunch of cables here over a dozen years ago. I’ve certainly done it a number of times. Many sound/music engineers have and many installation engineers in numerous different fields do so regularly. Nick’s results were entirely in line not only with theoretical predictions but also with everyone else‘s measurements, differences in the thousandths of a dB range and occasionally into the hundredths of a dB but always “way, way below of what’s audible”. Cable performance is extremely well known both theoretically and in practice and has been for many, many years. Our modern world simply wouldn’t exist if it were not the case.

I could do it myself in a weekend but I think it's a massive waste of time …

Doing something that millions of engineers and scientists have been doing routinely for a century and a half is obviously a massive waste of time. However, doing direct measurements, null tests and/or DBT/ABX is something I would recommend to many/most. Not because they might discover something not already well known but simply to create a realistic correlation between abstract scales/values (such as Hertz or dB) and their experience/hearing perception. If you already have a “realistic correlation” there‘s nothing to gain and it would be “a massive waste of time” but clearly many audiophiles have at least an unrealistic and commonly a ridiculously unrealistic correlation, apparently based on nothing but marketing and/or guessing, as evidenced by the number willing to believe claims of signals/differences being audible that are so ludicrously tiny they can’t even exist as sound waves.

TBH, despite his promise to come back, I think we’ve probably seen the last of @WaveTheory, in fact I was surprised he turned up in the first place. Presumably he thought misapplying/misrepresenting some cherry picked quotes from scientific sources/evidence (that he claimed did not exist in his video) would be enough and indeed in many audiophile forums it would probably be more than enough but not here or in other forums equally intolerant of the same tired old audiophile BS. I can’t see why he would come back, what could he do/say? Besides actually admitting his assertions were false (which presumably he won’t), what can he do other than just repeat the same pseudoscience BS (misrepresenting cherry picked “scientific” quotes) which he’s already learned doesn’t work here?

G

 

[MrHaelscheir]

@VNandor The point of my taking acoustic measurements of the entire system was mainly to target claims about cables doing such things as "smoothing treble peaks" despite the cable itself measuring flat, and because this method was the best that was within my means at the time. As gregorio said, there are already plenty of measurements directly measuring the cable independent of the transducer. Otherwise, I agree that this measurement setup having the signal-generating DAC/amp and the ADC on separate devices and USB connections is quite flawed for phase measurements where consistent alignment between signal and sample is essential, so yes, my speaker and HE-1 phase measurements should be taken with a grain of salt until I get to redo them with my new MOTU M2 audio interface. Given that, as I will show below, when using the MOTU M2 as a sample-synchronized unit driving the target amp with its DAC and synchronizing that with a sample from the ADC on the same device, phase response measurement consistency is drastically improved, whereby indeed, the cables make absolutely no difference to either the magnitude or phase response. So all my previous ramblings about phase being extremely sensitive to the tiniest pad perturbations between measurements were invalid, though it probably does still stand that what tiny earpad-related phase response changes that may occur would be orders of magnitude above any introduced by the cables themselves.

New MOTU M2 audio interface for better phase response measurements and measurement averaging:

Here we have this monstrous connection of the in-ear mics to the MOTU M2:

As I had learned in https://www.audiosciencereview.com/...enser-microphones-with-motu-m2-and-rew.49384/, the microphones I was using in https://www.head-fi.org/threads/mez...eadphone-official-thread.959445/post-17743502 (post #5,152) required 3-5 V "plug-in power" (these specific mics are rated from 2 V to 10 V), whereby I needed to convert the MOTU M2's 48V phantom power to that, its having been likely that the Axagon ADA-17 I was originally using already supplied that plug-in power, though I don't see that explicitly documented. I opted for the RØDE VXLR Pro both because it unlike the VXLR+ converts the unbalanced input into a balanced signal, and because the amazon.ca cost conversion from the amazon.com price was more favourable at the time. The in-ear microphones were coalesced into a single TRS stereo jack for the convenience of the original Earfish application of HRTF measurement with the Axagon ADA-17 strapped to your head. This is run into a 3.5 mm female TRS to dual mono 1/4" male TS adapter, my then using adapters to connect those to the VXLR Pros' 3.5 mm TS inputs. I couldn't find direct 3.5 mm female to dual mono 3.5 mm male adapters, at least not of the length I needed.

Don't mind the balanced XLR headphone switchbox and Genelec volume knob (first, you can see in addition to my existing posts across this forum that I do have these cables; for the actual measurements, the switchbox was removed so I could rest a hand on top of the MOTU M2 to help reduce the noise floor, and the cables were plugged directly into the FiiO K9 Pro ESS. On the right at the back of the MOTU M2, I have a dual male 1/4" TRS to male 4.4 mm Pentaconn adapter running from the MOTU M2's monitor/line out to the FiiO K9 Pro ESS's balanced line in, the K9 Pro ESS being set to bypass the DAC and feed that line input to the attenuator and THX 788+ amp.

In REW, the MOTU's inputs and outputs are selected, yielding sample-synchronized measurements with the aforementioned excellent phase response consistency.

Results:

Other than the above, all the same methods outlined in https://www.head-fi.org/threads/in-...c-headphones-frs-are-indeed-identical.970202/ were followed. For volume-matching, the unEQed Meze Elite was driven with REW's "Check levels" signal such that an SPL meter with a wind shield foam separating the mic capsule to be around 2 cm from the Meze Elite's left channel's grille on an unsealed cup held in one's hand registered around 100 dBA, where MOTU M2's preamp gain knobs then being adjusted such that REW (after disabling input volume control or setting it to 1.00) registers a "Check levels" value of 100 dB SPL zero-weighted for both channels. The HiFiMan Arya Stealth was set to register the same by adjusting the FiiO K9 Pro ESS's volume knob while keeping the gain knobs at the same level.

The full measurements to be viewed on REW as well as the text export of the magnitude and phase responses can be found at https://drive.google.com/file/d/1u9MkRnZZ8CUYX8IJH9pdKH_AZXBe39H-/view?usp=drive_link. Each measurement in the REW file has a description and timestamp. Clicking between measurements within REW allows for easier comparison of differences or the lack thereof. Both 4M and 512k length measurements were taken as a demonstration of the magnitude and phase response consistency between measurements of the same headphone, channel, and cable.

Meze Elite phase response measurements:
Frequency response for Meze Elite hybrid pads with Meze OFC 99.95% Copper Standard Cable Balanced 4 pin XLR Cable:

Frequency response for Meze Elite hybrid pads with Meze Mini XLR to 4.4 mm Balanced Silver Plated PCUHD Copper Premium Cable:

HiFiMan Arya Stealth phase response measurements:

Note that I have shown here the 512k-length instead of 4M-length measurements since the latter exhibited more timing shift between the two measurements, a shift that could occur with the same cable.

Frequency response for HiFiMan Arya Stealth with fabric sleeve stock (crystalline silver and/or copper) cable, unbalanced and volume-matched:

Frequency response for HiFiMan Arya Stealth with Meze Mono 3.5 mm to 4.4 mm Balanced Silver-Plated Copper Upgrade Cable:

As can be seen, the magnitude and phase responses between cables for the same headphone, even for unbalanced on medium instead of low gain, are effectively identical.

Finally, below is an example of the magnitude and phase response changes resulting from pressing hard on the left cup, particularly the upper midrange getting elevated:

 

[gregorio]

The point of my taking acoustic measurements of the entire system was mainly to target claims about cables doing such things as "smoothing treble peaks" despite the cable itself measuring flat, and because this method was the best that was within my means at the time.

TBH, I’m not really sure of “the point of you taking acoustic measurements“. There are plenty of measurements of cables themselves and as we’re talking about analogue signal cables then if there were “such things as smoothing treble peaks” then we would see/measure treble peaks in the analogue signal being smoothed/altered, which we don’t. Taking acoustic measurements invariably introduces other variables, transducers, etc., and if there were altered treble peaks or anything else, when there aren’t any (of any audible significance) in the analogue signals exiting the cables, then obviously that must be caused by something downstream and not the cable. So taking acoustic measurements is just confusing the issue and potentially measuring some component/effect other than cable performance.

Are you taking acoustic measurements because the claims are of “sound quality” improvements/differences and therefore you’re measuring the sound? If some audiophiles do not understand or accept that analogue audio cables do not transfer sound, they transfer analogue audio signals, or do not understand or accept what an analogue audio signal is, then is there any reason to think they would understand or accept the sound measurements you have presented?

G

 

[bigshot]

a lot of work just to prove something that is already self-evident... it seems the purpose of the test is testing for testing's sake, not finding a result.

 

[MrHaelscheir]

TBH, I’m not really sure of “the point of you taking acoustic measurements“. There are plenty of measurements of cables themselves and as we’re talking about analogue signal cables then if there were “such things as smoothing treble peaks” then we would see/measure treble peaks in the analogue signal being smoothed/altered, which we don’t. Taking acoustic measurements invariably introduces other variables, transducers, etc., and if there were altered treble peaks or anything else, when there aren’t any (of any audible significance) in the analogue signals exiting the cables, then obviously that must be caused by something downstream and not the cable. So taking acoustic measurements is just confusing the issue and potentially measuring some component/effect other than cable performance.

Are you taking acoustic measurements because the claims are of “sound quality” improvements/differences and therefore you’re measuring the sound? If some audiophiles do not understand or accept that analogue audio cables do not transfer sound, they transfer analogue audio signals, or do not understand or accept what an analogue audio signal is, then is there any reason to think they would understand or accept the sound measurements you have presented?

G

Of course, the "treble peaks" (which I interpret as regarding the transfer function of the system rather than a frequency decomposition of the source signal being passed through said transfer function) do not exist in the cables, whereby I am using acoustic measurements to refute the claim that cables can magically alter a transducer's performance to possess (as when describing a cable as being "peaky", which is a damnable offense) or smooth out the peaks inherent to the transducer. Ascribing the property of cables applying dynamic compression to treble content in the signal is a different story... And if one genuinely had no idea that the peaks could be originating from the transducer and its interaction with one's ears rather than anywhere else, they are quite unfortunately mistaken. And an adamant refusal to understand what "sound" actually is cannot be helped.

@bigshot Of course, testing of the cables' own transfer function has already been extensively analyzed. I am simply presenting another form of data proving that those findings indeed do not change after you have added the transducer and human ears into the mix, removing any doubt as to whether a cable could somehow induce changes after the signal has "left" its terminal.

 

[bigshot]

It seems to me that if a century of scientific study and proof isn't enough to convince a died in the wool audiophool, one more test isn't going to convince them either. We spend a disproportionate amount of time here in Sound Science debating things that just aren't debatable. I can think of things that would be better to spend time on than whether cables can sound like something.