Milind Kunchur

[dc655321]

That old chestnut catch-all

I did start reading but several serious technically flaws meant I did not continue.

Serious flaws such as...?

The signal pulled out of the noise in this paper seems much too high to be believable, imo. But I would need much more study than my initial skimming to be certain, let alone determine the "why" of his results.

The casual dismissal by you and others here is disappointing. If you have the chops to critique the paper, be explicit ffs!

 

[Frank Dernie]

Serious flaws such as...?

The signal pulled out of the noise in this paper seems much too high to be believable, imo. But I would need much more study than my initial skimming to be certain, let alone determine the "why" of his results.

The casual dismissal by you and others here is disappointing. If you have the chops to critique the paper, be explicit ffs!

"
HEA systems require meticulous setup and attention to detail (e.g., precise speaker positioning for best acoustic coupling with the room, use of spikes to prevent speaker recoil, etc.)"


in the second paragraph was enough to know he doesn't know his shit.

 

[SIY]

Serious flaws such as...?

The signal pulled out of the noise in this paper seems much too high to be believable, imo. But I would need much more study than my initial skimming to be certain, let alone determine the "why" of his results.

The casual dismissal by you and others here is disappointing. If you have the chops to critique the paper, be explicit ffs!

Untangling statistical sleight of hand takes serious effort, and the way it’s done here suggests that it’s needed to show any significance. Given Kunchur’s track record and demonstrated unwillingness to admit error, it’s highly unlikely to be worth the effort, and casual dismissal is appropriate.

If Stan Pons published a paper claiming First Law violations, I would have the same reaction. Life’s too short to waste it on incompetents.

 

[dc655321]

Untangling statistical sleight of hand takes serious effort, and the way it’s done here suggests that it’s needed to show any significance.

Yes.
My concern is that papers like this seem to live on largely unchecked in any official way by authoritative reviewers. I'm thinking of Stuart's MQA paper on "time smearing" (?!?) as another example. These results will become fodder (or gospel?) for those who need it, and imo, ought to be professionally debunked before they become rooted as audiophool lore.

I understand you think this guy needs to stay in his lane and has a history of bad (audio) science, but I disagree that casual dismissal is adequate.
Maybe I'm getting ahead of myself, and a proper critique will be forthcoming. I do hope so.

 

[SIY]

Yes.
My concern is that papers like this seem to live on largely unchecked in any official way by authoritative reviewers. I'm thinking of Stuart's MQA paper on "time smearing" (?!?) as another example. These results will become fodder (or gospel?) for those who need it, and imo, ought to be professionally debunked before they become rooted as audiophool lore.

I understand you think this guy needs to stay in his lane and has a history of bad (audio) science, but I disagree that casual dismissal is adequate.
Maybe I'm getting ahead of myself, and a proper critique will be forthcoming. I do hope so.

Wrong stuff, even when thoroughly critiqued, manages to live on forever in the fashion audio swamp world. Look at Oohashi and Hawksford’s Essex Echo, which get dragged out on the regular. Ditto Kunchur’s incompetent work on timing. Or, in my analogy, the stream of stuff from people like Rothwell on cold fusion.

It’s a sad fact of life that there’s an audience for nonsense, and they cannot be dissuaded. When authors dig in and refuse to withdraw their papers, there’s not much that can be done.

 

[dc655321]

When authors dig in and refuse to withdraw their papers, there’s not much that can be done.

That's a fair statement.
Ideally, publications' peer review process would catch and call out any mistakes and nonsense.
The JAES does seem to struggle with that...

 

[okok]

milan kundera?

 

[LTig]

Serious flaws such as...?

He says that if there is an audible difference then the electrical signal must be different (which I'm fine with) and then he measures the sound in the room with REW??? Just connect an AP to the end of the cable and measure it, this gives a result with a precision of many (!) orders higher.

 

[dc655321]

He says that if there is an audible difference then the electrical signal must be different (which I'm fine with) and then he measures the sound in the room with REW??? Just connect an AP to the end of the cable and measure it, this gives a result with a precision of many (!) orders higher.

AFAIU, in the secion of the paper you're referring to, the author was measuring the acoustic characteristics of the listening room.
How might one do that looking at electrical signals?

A more interesting question may be, What bearing would such a measurement have on detection thresholds?
The author does not address this, and it's likely not relevant (i.e. acoustic measurement uncertainty would swamp electrical measurement uncertainty anyway!).

 

[Blumlein 88]

Well the measurements for cables suffer from the old difference is all way ultrasonic. The noise figures seem high, but include the high bandwidth. He says it matters because it may effect the following amp which he never does an example of.

So next is his listening tests which I've not read yet.

 

[dc655321]

So next is his listening tests which I've not read yet.

It's a bit, um, unorthodox - match a pre-determined description of sonic attributes to what is heard.

The subjects’ task was to match the aforementioned standardized subjective adjectives to determine the order (A/B or B/A) of each random pair.

 

[Blumlein 88]

It's a bit, um, unorthodox - match a pre-determined description of sonic attributes to what is heard.

What kind of results statistically speaking?

 

[LTig]

AFAIU, in the secion of the paper you're referring to, the author was measuring the acoustic characteristics of the listening room.
How might one do that looking at electrical signals?

A more interesting question may be, What bearing would such a measurement have on detection thresholds?
The author does not address this, and it's likely not relevant (i.e. acoustic measurement uncertainty would swamp electrical measurement uncertainty anyway!).

I refer to chapter 4 in the AES "paper":

4 ELECTRICAL CHARACTERIZATION​

If two audio configurations are audibly distinguishable, then physical differences between their signals must necessarily exist. ​

OK, I did interpret the word physical as meaning electrical. Can't hide the fact that I'm an EE ...

However, measuring them and interpreting their relevance for audio performance is challenging because of the difficulties in matching the extraordinary capabilities of the ear, and because of our incomplete understanding of auditory neurophysiology. For example, the ear has a 120 dB dynamic range and the sensitivity to detect a cochlear basilar-membrane amplitude of ~1 pm [39]–[41]. It was not the primary goal of the present work to pinpoint the exact physical reason/s for why the interconnects sound different, just that they do. Nevertheless, in the electrical measurements presented below, the sonically superior higher-end interconnect did perform better in all of these measurements.​

And this is the stupid part. Instead of measuring the audio signal with a microphone it would have been sufficient to measure the electrical signal at the end of the cable. Then you know the real difference and can calculate how much difference that is in SPL at the listening position.

 

[dc655321]

What kind of results statistically speaking?

43/59 correct, with an unbelievable, p < 0.0005.
I am not familiar with the Wilcoxon signed-ranks test used here.

 

[dc655321]

I refer to chapter 4 in the AES "paper":

4 ELECTRICAL CHARACTERIZATION​

If two audio configurations are audibly distinguishable, then physical differences between their signals must necessarily exist. ​

OK, I did interpret the word physical as meaning electrical. Can't hide the fact that I'm an EE ...

However, measuring them and interpreting their relevance for audio performance is challenging because of the difficulties in matching the extraordinary capabilities of the ear, and because of our incomplete understanding of auditory neurophysiology. For example, the ear has a 120 dB dynamic range and the sensitivity to detect a cochlear basilar-membrane amplitude of ~1 pm [39]–[41]. It was not the primary goal of the present work to pinpoint the exact physical reason/s for why the interconnects sound different, just that they do. Nevertheless, in the electrical measurements presented below, the sonically superior higher-end interconnect did perform better in all of these measurements.​

And this is the stupid part. Instead of measuring the audio signal with a microphone it would have been sufficient to measure the electrical signal at the end of the cable. Then you know the real difference and can calculate how much difference that is in SPL at the listening position.

If one was establishing a lack of credibility in a court of law, sure, there is some material here and in his past paper.
But, this does little to poke holes in the incredible results of this paper - you're just pointing out an alternative (and correct) way of establishing that the signals would differ.

As you quoted, "It was not the primary goal of the present work to pinpoint the exact physical reason/s for why the interconnects sound different, just that they do."

The author has found a deafeningly loud statistical signal where there ought to be a nano-whisper.
Why? What mistakes were made?

Poking fun at his ineptitude(s) is entertaining, but not a very useful refute.

 

[j_j]

Oh boy. First his test does not have any controls built in. There are neither A vs. A nor "known audible" stimuli included.

The way the long term test switched cables is not detailed, either, nor is the actual blinding process that I could find.

Finally, his description of his listening room sounds like an ITU room, than a typical listening setup.

I would need more information, in particular gain and noise immunity of his single-ended inputs vs. Pin 1 problems, before I would make any other comments.

But the 40 or 50 seconds during which he would have to swap cables could allow for a terrifying amount of subject-experimenter leakage as well. More detail is required.

 

[dc655321]

But the 40 or 50 seconds during which he would have to swap cables could allow for a terrifying amount of subject-experimenter leakage as well. More detail is required.

Thanks @j_j for weighing in.
Sounds like no cable swapping required though:

They were both continuously connected to a class-A solid-state Spectral DMA-250S Studio Universal Amplifier, which has balanced and single-ended switch-selectable inputs. This arrangement avoids the need to disconnect and reconnect interconnects during trials, and avoids intervening external switch boxes and additional cables and circuit paths, which might compromise fidelity.

 

[j_j]

Thanks @j_j for weighing in.
Sounds like no cable swapping required though:

That still does not explain the blinding methods. Was this double-blind, i.e. was the person who flipped the switch not only not in the room when flipping, but also not in the room during the listening? You'd be astonished at what a subject will unconsciously use for distinctions.

Also, to what extent was the through-amplifier gain tested in both positions. This may have been answered but I'm at work and kind of swamped.

Matching is best if to .1dB or better. .2dB gain difference is more than enough to cause extremely strong preferences.

 

[dc655321]

I'm at work and kind of swamped.

Bummer. Me too. Day jobs suck sometimes!
Hopefully you have time and interest to weigh in again when you can devote more attention to the paper.

 

[scott wurcer]

Recent work proved, through psychoacoustic testing, that analog interconnect cables supplying line-level signals to an amplifier can discernibly affect the sound quality [1]

First line of abstract references an unpublished self reference as proof, why go any further?