Community

AES Journal Forum

Cable Pathways Between Audio Components Can Affect Perceived Sound Quality

(Subscribe to this discussion)

Document Thumbnail

The arena of highest fidelity in music reproduction, sometimes referred to as high-end audio, has many controversial claims and contentious issues. One such controversy is whether the cables and topology used to interlink components together make an audible difference. There seems to be a disparity between anecdotal experiences reported by audiophiles and published formal scientific research as to theminimal changes in system configuration that can be audibly distinguished. With the motivation of bridging this divide—which may originate from differences in instrumentation and subject-listening conditions used by the two groups—this work utilized a high-performance audio system and extended-duration listening protocol that more closely resembles audiophile auditioning conditions. With these measures the present work was able to prove through direct psychoacoustic testing that two different analog-interconnect pathways can be audibly distinguished.

Author:
Affiliation:
JAES Volume 69 Issue 6 pp. 398-409; June 2021
Publication Date:

Click to purchase paper as a non-member or you can login as an AES member to see more options.

(Comment on this paper)

Comments on this paper

Johannes Didden
Johannes Didden


Comment posted June 7, 2021 @ 15:58:31 UTC (Comment permalink)

I would like to congratulate Professor Kunchur with a very interesting and thorough research. The subject has been discussed, on and off, for more than half a century now and studies like this are few and far between.

 
I do have one question though, which concerns the measurement of the frequency response of the tested cables, results of which are presented in fig. 3. What were the source- and load impedances on the cables used in this measurement?

Ammar Jadusingh
Ammar Jadusingh


Comment posted June 8, 2021 @ 15:47:30 UTC (Comment permalink)

So called "High End" aka "High Prices" Audio is a devotion to archaic stereo reproduction.

The concept of "Fidelity" there to what exactly, is somewhat questionable. Dr Floyd Tooles "Circle of confusion".
Worsened by the fact that 2 channels can't possibly reconstruct a real soundfield.
aes.org/e-lib/browse.cfm?elib=9136
Unclear how (possibly) detecting differnces in balanced/unbalanced cabling in a system helps here.

Roland Jacques
Roland Jacques


Comment posted June 9, 2021 @ 15:30:40 UTC (Comment permalink)

@Ammar Jadusingh: I'm afraid that the vast majority of audio producers and consumers only cares about the emotion of the music, and has little interest in physically correct wave fields. Quite on the contrary, many smart devices and smart speakers only deliver some vague spaciousness, or outright plain mono, often with compromised audio quality, along with a suboptimal listening environment. In comparison, a high-quality stereo reproduction is a revelation in spatial, tonal and dynamic accuracy. But I totally agree that especially in the cable business massive overpricing can be found, along with unscientific claims - therefore I applaud any solid, unbiased investigation into such claims, in order to support honest manufacturers who offer real, verifiable value.


Roland Jacques
Roland Jacques


Comment posted June 8, 2021 @ 15:47:39 UTC (Comment permalink)

Dear Prof. Kunchur, thanks for this very interesting article. As you point out, this field has been a controversial one for decades, sometimes bordering on metaphysical claims and beliefs. Any rigorous scientific approach is therefore highly appreciated, and very helpful for consumers, manufacturers and researchers alike. Upon studying your article, I had several thoughts:

First, regarding the choice of music program. As could be expected by the age of this particular recording, there is some rumble and hiss present, a quick analysis showed an SNR of only 40dB at peak moments, and only 17dB at soft moments. So it seems that the program noise was higher than any noise introduced by the cables, which leads to the question of the influence of the latter. A recording with less intrinsic noise might be better suited for the topic at hand. Also, a recording with stronger transient elements such as percussion or voice might be useful for analysis of the time-domain effects you mentioned. I would also suggest to, for some stimuli, increase the listening level above the 60dBA SPL average reported here.
 
Regarding the somewhat astounding noise floor figures: It seems to me that the reasons for this can be found in the setup as described below Fig. 4: Firstly, the 1 MOhm coupling probably makes the cables more susceptible to "antenna-like" EM induction than in the case of 10kOhm input impedance of the amplifier. Secondly, it is not surprising that the internal differencing applied to the +/- wires of cable A leads to effectively much lower noise, since this is the very goal of such a balanced topology (common mode rejection). Thirdly and most importantly, as you mention yourself, most of the noise is far outside of the audible spectrum; I'm not sure I can agree to the claimed "audible signature" due to "rectification-demodulation" of RF noise (I presume the input circuit of the amplifier is much more of a lowpass than a rectifier); most certainly not to the extent implied by your statement of 25.1dB effective SNR with the music program used here, which would be stupendous for a cable. I would suggest additional measurements to clarify these effects and their causes.
 
I also appreciate the thought which you put into the listening test design, and I can somewhat understand the concerns of fatigue and mental "overlap" when switching between the configurations too frequently, and the potential benefit of allowing time for building up a mental "image" of the music and its sonic properties. On the other hand, my experience in listening test design has taught me the very strong influence of the quickly diminishing auditory memory, combined with unavoidable fluctuations in the subject's mental and biological state. Therefore, the choice to have >5 minutes long segments, only 1 presentation per stimulus, and a pause of 40sec between them seems quite extreme to me, and I'm wondering whether the mental images of the sonic signature are really detailed and persistent enough to survive the quick decay of the auditory memory. This question alone would warrant a master thesis of its own, I think, and also allow for larger sample sizes and larger sets of stimuli.
 
I look forward to your insights on this, and of course to any further research in this quite relevant field. Thank you and best regards!

Default Avatar
James Hipperson


Comment posted June 10, 2021 @ 15:46:44 UTC (Comment permalink)

 "This DAC has two isolated buffered pairs of analog outputs — single-ended RCA and balanced XLR"

This to me sounds like a severe confounding variable in the experiment, unless it was the explicit intention to test balanced vs. unbalanced, which is not clear from the rest of the paper. 

The abstract says:

"the present work was able to prove through direct psychoacoustic testing that two different analog-interconnect pathways can be audibly distinguished."

I think a fairer and more specific summary of the experiment, was that it was demonstrated that listeners can distinguish between balanced and unbalanced interconnections, which is interesting in itself despite being less general than implied here. 


Default Avatar
Author Response
Milind N. Kunchur


Comment posted June 13, 2021 @ 15:55:08 UTC (Comment permalink)

Hello everyone and thanks for your comments and interest. I'll try to answer the various questions to the extent that I have something useful to add. The answers to some comments are embedded somewhere in the long paper (sometimes in the footnotes). I must admit that I don’t check the forum regularly. On a related note, there will be a roundtable discussion of the paper on 10am EDT on June 29th arranged by JAES.

 *  The source and load impedances on the cables used in the REW  frequency-response measurements were 1 Ohm and 4300 Ohms respectively. I have ongoing research in which I am doing an in-depth electrical study of analog interconnects (mainly investigating uncommon time-domain distortions and their possible relationship to auditory neurophysiology). Here I am using various stand-alone signal generators and advanced oscilloscopes. The previous conclusions about frequency response remain unchanged with 100 ohms and 10 k-Ohms source and load impedances. I hope to finish this new work and paper by the end of the year.

* A properly setup 2-channel HEA system (something extremely rare) can produce a shockingly life-like 3-D representation, especially for small acoustic musical ensembles. Please see "3D imaging in two-channel stereo sound: portrayal of elevation", M. N. Kunchur, Applied Acoustics 175, 107811 (2021). A preprint can be downloaded from:
http://boson.physics.sc.edu/~kunchur//papers/Stereo-height—Kunchur.pdf

* The comments on optimizing the durations, gaps, etc. in the blind testing are good observations. We have a long way to go to find the most sensitive way to conduct blind testing. This work represents one step and improvement from past approaches that led to a detectable difference. But the questions about "quick decay of the auditory memory" are deep ones, which are perhaps best discussed at the roundtable (on 10am EDT on June 29th arranged by JAES). The 1 M-Ohm will certainly exacerbate noise pickup, which is why it was chosen — to amplify differences. However, noise was also measured at the amp output, with the cables terminated with the actual impedances in normal use.

* The various comments about balanced versus unbalanced are actually fully addressed in the paper. The wording of the conclusions and title state that "cable pathways" (which includes topology as well as the cables) and not just "cables" or just "topology" (i.e., balanced versus unbalance) produce an audible difference. As stated in the paper, to my knowledge, this may represent the smallest change in an audio system proven to be discernable through IRB approved blind listening tests.

 


Default Avatar
Author Response
Milind N. Kunchur


Comment posted June 23, 2021 @ 20:28:31 UTC (Comment permalink)

The link to the free downloadable preprint in my previous communicatin has not reproduced correctly and does not work. It should read (two simple dashes before Kunchur.pdf): 

http://boson.physics.sc.edu/~kunchur//papers/Stereo-height—Kunchur.pdf

Also only a portion of the URL got hot-linked.

This is for the paper "M. N. Kunchur, Applied Acoustics 175, 107811, 2021".

Thanks for your attention. 


Scott Dorsey
Scott Dorsey


Comment posted June 23, 2021 @ 20:28:44 UTC (Comment permalink)

1 ohm and 4300 ohm would seem to me to minimize differences in cable reactances rather than to exaggerate them.  Much home audio equipment has far higher output and input impedances which one would expect to cause greater frequency response problems with reactive cables.  And of course many commercially-made high end cables are deliberately made to be very very reactive in order to act as  a tone control.  So I would expect typical effects in the real world to be even greater than described in this paper.


Default Avatar
Eric Wenocur


Comment posted July 7, 2021 @ 15:36:57 UTC (Comment permalink)

The relative merits of "high-end audio" products have been debated for decades because most of the claims made for outrageously overpriced cables and power cords are indefensible. At the same time, this paper's claim that interconnect differences are audible is hardly news to anyone who works with audio equipment!

So was the point of the paper to show that a cable costing $1000/m is, in fact, better than one costing $25/m? Or simply to find the smallest possible change in "sonic character" that a human can perceive? If the former, one must ask if the purported difference is even due to the cables, since many other factors in the testing setup could cause a subtle difference in perception. The fact of comparing balanced and unbalanced connections is only the most obvious. The circuits comprising those different interfaces would also be somewhat different. The electrical characteristics and condition of the connectors might factor in. Measured differences of hundredths of a millivolt (that's .00001V) could be attributable to all sorts of effects. Even the selection of source material could skew the listening results in both physiological and psychological ways.

If the latter, one could devise much simpler (and less contentious) ways of trying to assess what listeners can or cannot perceive. But even so, attempting to quantify extremely subtle differences in sonic character, which is entirely subjective, prone to human error, and can only be described with adjectives that are also subjective, seems highly fraught.

This paper shows a valiant effort to control every aspect of a test scenario, and must have been an immense headache to perform. I found it painful to contemplate. But in the end I don't see how meaningful results are possible. More importantly, the test cannot answer two critical questions: What is sonic character or "sound quality"? And is it worth spending a boatload of money on any part of a system that *might* result in a miniscule change (compared to what?) in that quality? Unfortunately the consumers taken in by certain types of "audiophile" products are naive to the point of self-deception. But the effects of marketing, ignorance and ego is a subject for a different paper.


Subscribe to this discussion

RSS Feed To be notified of new comments on this paper you can subscribe to this RSS feed. Forum users should login to see additional options.

Join this discussion!

If you would like to contribute to the discussion about this paper and are an AES member then you can login here:
Username:
Password:

If you are not yet an AES member and have something important to say about this paper then we urge you to join the AES today and make your voice heard. You can join online today by clicking here.

AES - Audio Engineering Society