Objectivist or Subjectivist? Give Me a Break

[microstrip]

My god, how many ways do I have to say this?

If you'll read the thread, micro, you'll know that I only provided that link to show that there is data out there. I didn't endorse that particular study. I didn't even read it, and I don't intend to. (...)

Tim

Exactly my point. Your answer shows how seriously we should take your posts. You supply something you found in 30 seconds ( a student report, not a research paper) and concluded that there is surely a lot of relevant data that the others should search.

It's out there . Here's the very first reasonably objective thing I came across in a Google search, from a research paper done at MIT:

http://dspace.mit.edu/bitstream/handle/1721.1/46225/41567257.pdf

Plenty of data in there. Pretty convincing. And it took about 30 seconds to find. There's plenty more out there, just be sure to watch the sources carefully. White papers from high-end cable companies pop up a lot. Obviously they're not objective.

Tim

BTW, I notice you did not address any of the technical questions I raised, and focused only in this fait divers about so called research papers. Do you really dislike objective matters?

 

[Ethan Winer]

I am as skeptical of the viewpoint that "all cables sound the same" as I am of the viewpoint that there are huge differences between two well-designed and well-built cables. The difference is that the first claims to be able to make convincing arguments and present convincing data in its support, which I have yet to see.

What specifically would convince you that two competent cables sound the same? If someone blind tested you 20 times in a row and you got it right only half the time, would that be enough? What about measurements? If someone showed you measurements of frequency response, noise, and distortion where all differences were 100 dB down, would that do it? BTW, this is a serious question, not meant to be combative. I honestly would like to know what exactly it takes to convince someone who is otherwise sitting on the fence.

--Ethan

 

[Phelonious Ponk]

Exactly my point. Your answer shows how seriously we should take your posts. You supply something you found in 30 seconds ( a student report, not a research paper) and concluded that there is surely a lot of relevant data that the others should search.



BTW, I notice you did not address any of the technical questions I raised, and focused only in this fait divers about so called research papers. Do you really dislike objective matters?

One more time, and then you can take your inability to understand what's being said elsewhere. I provided the link only to demonstrate that there is, indeed data in existence regarding audio cables. I skimmed it only enough to know that it compared measurements of noise levels, impedance, etc. I've been repeatedly clear about that, and micro, you can take that as seriously as you like, I really don't give a damn. You are now going on ignore before I lose my temper and break the rules of the forum.

Tim

 

[rbbert]

What specifically would convince you that two competent cables sound the same? If someone blind tested you 20 times in a row and you got it right only half the time, would that be enough? What about measurements? If someone showed you measurements of frequency response, noise, and distortion where all differences were 100 dB down, would that do it? BTW, this is a serious question, not meant to be combative. I honestly would like to know what exactly it takes to convince someone who is otherwise sitting on the fence.

--Ethan

Well, the first thing I'd like to see is a study design, perhaps using a passive preamp, comparing 3' and 30' lengths of the same cable where there is a measurable difference between the two in the audio range (where it should be detected). Since I haven't even read about that, I'm unwilling to speculate yet about what would be nice after that demonstration.

 

[JackD201]

What specifically would convince you that two competent cables sound the same? If someone blind tested you 20 times in a row and you got it right only half the time, would that be enough? What about measurements? If someone showed you measurements of frequency response, noise, and distortion where all differences were 100 dB down, would that do it? BTW, this is a serious question, not meant to be combative. I honestly would like to know what exactly it takes to convince someone who is otherwise sitting on the fence.

--Ethan

A sample size of 100% confidence if you mean 2 competent cables that sound the same to EVERYBODY. You can keep the hearing impaired and little children in to pad your assertion.

Of course it would be dandy if someone actually defined the parameters of "competent" first. Throw in the ranges of relevant external environmental influences as well. What's a test without a control?

 

[j_j]

Is this true? I believe that there is a correlation that has stood the test of time, and that is the one that says that simple linearity between input and output sounds best.

No.

While it hasn't been published much, almost because it seems to trigger near-violence in some quarters, there is a body of knowledge about euphonic distortions.

Guitar amplifiers are an example, but of course you can argue that they are musical instruments themselves, not just reproduction systems.

Let me ask you, then.

Let us hypothesize for a minute a system that exhibits .1% distortion (not counting noise) at -20dB, 1% at -10dB, and 10% at 0dB relative to its overload point.

Now, what will happen with the harmonic splatter that happens when the system gets near peak levels?

In particular, what will happen with the relationship between loudness (which is a formally defined term that means sensation level) and intensity (we're talking electronic here, not acoustic intensity) when this system starts to get near overload and the THD (which is relative to the input level, not an absolute least-mean-squares measurement) rises rapidly?

Now, assuming you correctly understand that the loudness increases disproportionately to signal level due to the bandwidth expansion, what does this do to the perceived dynamic range of the signal?

What does it mean, say, when the loudness of the signal is exaggerated by a factor of 2 (only on peaks) relative to a purely linear system?

What increase in signal level would provide the same loudness increase?

 

[j_j]

Exactly my point. Your answer shows how seriously we should take your posts. You supply something you found in 30 seconds ( a student report, not a research paper) and concluded that there is surely a lot of relevant data that the others should search.



BTW, I notice you did not address any of the technical questions I raised, and focused only in this fait divers about so called research papers. Do you really dislike objective matters?

You are really holding him to two different standards here, one before the question of a study was answered, and one after.

Do you disagree with the position that electrical performance is the sole feature of cable audibility in a blind test?

 

[j_j]

Well, the first thing I'd like to see is a study design, perhaps using a passive preamp, comparing 3' and 30' lengths of the same cable where there is a measurable difference between the two in the audio range (where it should be detected). Since I haven't even read about that, I'm unwilling to speculate yet about what would be nice after that demonstration.

If this is as far as you've gotten in test design, you haven't once noticed my comments about positive and negative controls in a test. You haven't explained why you would use a passive preamp (although that would exaggerate differences due to RLC), or how you might do clickless, seamless, at-will switching to enable ultimate test sensitivity.

So what are you arguing about here?

 

[j_j]

A sample size of 100% confidence if you mean 2 competent cables that sound the same to EVERYBODY. You can keep the hearing impaired and little children in to pad your assertion.

Of course it would be dandy if someone actually defined the parameters of "competent" first. Throw in the ranges of relevant external environmental influences as well. What's a test without a control?

You do realize that it is possible to measure the performance of a population without testing every individual, I hope.

Now, do you understand what 100% confidence even means, or how you might achieve that?

Your arguments suggest that you need to understand the mathematics of both testing of a population and how such testing would be evaluated, with particular attention to type 1 and type 2 error.

 

[JackD201]

Of course I do. Do you understand humor?

 

[Orb]

You are oversimplifying the problem. No system is completely linear, and IMHO the main issue is what are the really bad deviations, and how to minimize or mask them.

Signal handling is a compromise. Just measuring and trying to minimize the deviations without discernment - this means at less correlating them with perceived sound quality in my view - is not the solution. YMMV.

Indeed.
This thread is going like every other on here such as the one discussing do we measure all that is needed/do we interpret said data/do we have complete test process/etc.

Case in point on linearity; no-one has proposed a test-measurement-data analysis that truly looks at the linearity of actual sound through audio equipment; because one needs to also look at the interraction of all partials/harmonics in both amplitude/harmonic-partials/over time for complex note.
I know of several other engineer scientists with classical science backgrounds and worked in research who also mull over the idea about the way forward requires an incredibly complex procedure to do and a heck of a lot of maths as well; that being testing involving a synthesized polyphonic sound that also is mathematical in attack/decay and amplitude for multiple complex harmonics-partials (major chord but this would not be a traditional instrument but a digitally synthesised one).

Anywya how does one say speakers are linear when their speaker decays are so different even if they both have identical frequency response (lets assume two speakers have identical flat frequency response at 1khz but have real world very different decays).
More rhetorical than really wanting a response because it should have its own thread and one could argue around technical semantics and various points on that subject anyway.

Regarding one that does fit closer to this thread:
How does one measure linearity for the cumulative effects of dither or the various types of dither used as they are not all equal when comparing digital in and the digital processed out?
There is no quick and easy answer to this because only a few have done a test involving this and they noticed sound quality perception was worse as more dither was applied, good studio sound engineers do mention the application of dither is critical and notice sound differences but further research is required IMO (I am aware of the AES publications, and also from others).
One cannot rely upon frequency range/amplitude/a single sinewave-tone output to see effects (we are not talking about comparing non-dithered tone to a dithered one nor truncation or limited number of bits-resolution issues), another one that sort of falls into the cable discussion in that differences should be negligible and inaudible if one takes into consideration and deals with the issues I mention in brackets but yet it seems there are those with qualified experience suggesting it is audible - until better research and testing is done we will never know if it truly is or why.

Cheers
Orb

 

[Groucho]

No.

While it hasn't been published much, almost because it seems to trigger near-violence in some quarters, there is a body of knowledge about euphonic distortions.

Guitar amplifiers are an example, but of course you can argue that they are musical instruments themselves, not just reproduction systems.

Let me ask you, then.

Let us hypothesize for a minute a system that exhibits .1% distortion (not counting noise) at -20dB, 1% at -10dB, and 10% at 0dB relative to its overload point.

Now, what will happen with the harmonic splatter that happens when the system gets near peak levels?

In particular, what will happen with the relationship between loudness (which is a formally defined term that means sensation level) and intensity (we're talking electronic here, not acoustic intensity) when this system starts to get near overload and the THD (which is relative to the input level, not an absolute least-mean-squares measurement) rises rapidly?

Now, assuming you correctly understand that the loudness increases disproportionately to signal level due to the bandwidth expansion, what does this do to the perceived dynamic range of the signal?

What does it mean, say, when the loudness of the signal is exaggerated by a factor of 2 (only on peaks) relative to a purely linear system?

What increase in signal level would provide the same loudness increase?

Are you suggesting that amplifier clipping enhances a system's apparent dynamic range, because "harmonic splatter" fools us into thinking that our Fletcher Munsons are being stimulated?

One aspect that simply fails to make much sense in discussion of the deirability of certain forms of harmonic distortion, is that once there's more than one note playing, it ceases to be harmonic distortion at all. The amplifier doesn't break the sound down into individual sources, to each apply a bit of 2nd, a touch of 3rd, and a whiff whaff of 4th, then re-assemble the overall waveform for the speaker; it simply overlays a wobbly transfer function over the whole waveform, causing non-harmonic intermodulation distortion.

This is summed up very well by the Stereophile article on 'euphonic distortion':

Any device that introduces harmonic distortion on a sinusoidal input will also produce amplitude intermodulation distortion—sum and difference frequencies—on a complex input like a music signal. These intermod components are generally dissonant, and, what's more, they very quickly become the dominant component of the distortion as signal complexity increases. A seminal paper written during World War II by two British Post Office researchers investigated this and came to some startling conclusions (footnote 4). For instance, if the signal comprises 30 or more component frequencies, then the distortion power contained within the harmonic distortion products will be at least two orders of magnitude lower than (ie, less than 1% of) that contained within the intermodulation products. In such circumstances it is difficult to credit that the consonance or dissonance of the harmonic components can have any significant effect on the perceived quality of the distortion.

http://www.stereophile.com/reference/406howard/index.html

Maybe this is why people running vinyl/valve/full-range systems tend towards girl-and-guitar i.e. a small number of notes played at any one time. Any claimed preference for 'euphonic distortion' will be entirely dependent on the music played, and implicitly, the musical taste of the listener. This sort of thing must surely be a major problem in any claims for superiority of one piece of equipment over another based on listening tests..?

 

[microstrip]

(...)

Do you disagree with the position that electrical performance is the sole feature of cable audibility in a blind test?

J_J,

IMHO your statement is too imprecise and vague to be analyzed. Unless it is qualified, electrical performance can mean many different things in audio and we can have good and erroneous blind tests.

 

[microstrip]

(...)

Maybe this is why people running vinyl/valve/full-range systems tend towards girl-and-guitar i.e. a small number of notes played at any one time. Any claimed preference for 'euphonic distortion' will be entirely dependent on the music played, and implicitly, the musical taste of the listener. This sort of thing must surely be a major problem in any claims for superiority of one piece of equipment over another based on listening tests..?

Many papers have been published on this subject of of "harmonic splatter", the most interesting ones in the in the 50's, when power was limited and people were debating triodes versus pentodes. I read them mainly by curiosity, not professionally, in the british magazine Wireless World, as I did not have access to the US audio electronics literature of that period. Unfortunately it was all paper, and looking for them again is not now a fast job. IMHO the level of debate was much deeper than in Stereophile or any similar magazine.

I have to disagree with your statement about girl-and-guitar. Girl-and-guitar are usually nice in any system. IMHO valve (let us go on use the UK designation) typically sounds better in complex music, such as symphonic fortissimos or massed strings, or recordings needing to transmit the feeling of power and drama without sounding too loud. Many people think that chamber music is easy to reproduce, but again IMHO they are wrong - most instruments need a high dynamic range and the energy density is sometimes very high. Surely these are general feelings - it is always easy to find particular cases that reverse this feeling about SS versus valves.

BTW, we should also consider good opera recordings. I have seen Mirella Freni reducing a 200 pound Krell to smoke while singing Madama Butterfly ...

 

[Atmasphere]

Are you suggesting that amplifier clipping enhances a system's apparent dynamic range, because "harmonic splatter" fools us into thinking that our Fletcher Munsons are being stimulated?

Actually this is a well-known phenomena with amplifier designers. The ear uses odd ordered harmonics (specifically the 5th, 7th and 9th) to sort out how loud a sound is (IOW it does not gauge that from the fundamental tone) so if the system is contributing odd ordered harmonics even in trace amounts, it will tend to sound louder than in ought to. This is easily proven out with a sound pressure meter. FWIW *** Fletcher Munson curve is really showing us that the human ear is tuned to birdsong frequencies, and for a very good reason.

One aspect that simply fails to make much sense in discussion of the deirability of certain forms of harmonic distortion, is that once there's more than one note playing, it ceases to be harmonic distortion at all. The amplifier doesn't break the sound down into individual sources, to each apply a bit of 2nd, a touch of 3rd, and a whiff whaff of 4th, then re-assemble the overall waveform for the speaker; it simply overlays a wobbly transfer function over the whole waveform, causing non-harmonic intermodulation distortion.

This is summed up very well by the Stereophile article on 'euphonic distortion':

http://www.stereophile.com/reference/406howard/index.html

Maybe this is why people running vinyl/valve/full-range systems tend towards girl-and-guitar i.e. a small number of notes played at any one time. Any claimed preference for 'euphonic distortion' will be entirely dependent on the music played, and implicitly, the musical taste of the listener. This sort of thing must surely be a major problem in any claims for superiority of one piece of equipment over another based on listening tests..?

The lower ordered harmonics (2nd-4th) have been shown to be less critical to the human ear, contributing to what audiophiles call 'warmth', 'bloom', etc. This is because the ear translates distortion into tonality. It is why two circuits of different topology can have tonal differences between them, even though they measure flat on the bench. We simply are not measuring the right thing.

And FWIW, although I run vinyl and valves and my system is very full-range (although you may be referring to single-driver loudspeakers), I rarely listen to girl-and-guitar. That sounds like some sort of grossly inaccurate stereotype to me. IMO this comment can best be regarded as specious. I know a lot of audiophiles that play tubes and vinyl and they have very diverse tastes as one might expect.

 

[Phelonious Ponk]

Actually this is a well-known phenomena with amplifier designers. The ear uses odd ordered harmonics (specifically the 5th, 7th and 9th) to sort out how loud a sound is (IOW it does not gauge that from the fundamental tone) so if the system is contributing odd ordered harmonics even in trace amounts, it will tend to sound louder than in ought to. This is easily proven out with a sound pressure meter. FWIW *** Fletcher Munson curve is really showing us that the human ear is tuned to birdsong frequencies, and for a very good reason.


The lower ordered harmonics (2nd-4th) have been shown to be less critical to the human ear, contributing to what audiophiles call 'warmth', 'bloom', etc. This is because the ear translates distortion into tonality. It is why two circuits of different topology can have tonal differences between them, even though they measure flat on the bench. We simply are not measuring the right thing.

And FWIW, although I run vinyl and valves and my system is very full-range (although you may be referring to single-driver loudspeakers), I rarely listen to girl-and-guitar. That sounds like some sort of grossly inaccurate stereotype to me. IMO this comment can best be regarded as specious. I know a lot of audiophiles that play tubes and vinyl and they have very diverse tastes as one might expect.

I'm sure. I doubt that Groucho mean the description as any kind of absolute; what about the substance of his comment? IE:

The amplifier doesn't break the sound down into individual sources, to each apply a bit of 2nd, a touch of 3rd, and a whiff whaff of 4th, then re-assemble the overall waveform for the speaker; it simply overlays a wobbly transfer function over the whole waveform, causing non-harmonic intermodulation distortion.

Tim

 

[Atmasphere]

I'm sure. I doubt that Groucho mean the description as any kind of absolute; what about the substance of his comment? IE:

"
The amplifier doesn't break the sound down into individual sources, to each apply a bit of 2nd, a touch of 3rd, and a whiff whaff of 4th, then re-assemble the overall waveform for the speaker; it simply overlays a wobbly transfer function over the whole waveform, causing non-harmonic intermodulation distortion. "

Tim

Non-harmonic intermodulations are processed by the ear as tonality, typically brightness. Intermodulations can occur in a variety of manners, essentially anything that is non-linear in the system can cause intermodulations (transformers, transistors, pentodes, poor layout, poor topology choices, etc.)- I have seen a dirty switch make a big difference. One inharmonic intermodulation point occurs at the feedback node of any amplifier (Crowhurst), which is one reason I avoid using loop feedback. Interestingly, the ear has the ability to hear about 20 db into a noise floor of the right type (an exception to the masking rule, apparently) if that noise floor is similar to the sound of wind (I personally suspect this is a consequence of survival/evolutionary influences). If the noise floor is composed of harmonics, intermodulations and inharmonic artifacts, the ear cannot penetrate that floor and it will thus describe the lowest level of detail available. Zero feedback circuits can have a more towards a natural noise floor similar to wind; this may explain why they tend to have more detail even when the apparent noise floors measure at the same level. More research is needed IMO.

 

[Robh3606]

IMHO your statement is too imprecise and vague to be analyzed. Unless it is qualified, electrical performance can mean many different things in audio and we can have good and erroneous blind tests.

Hello Micro

How about Frequency Response as your only criteria?? What else would you add maybe phase and impulse response?? Worst case maybe measure 10 meter lengths for subwoofer duty??

Rob

 

[Phelonious Ponk]

Non-harmonic intermodulations are processed by the ear as tonality, typically brightness. Intermodulations can occur in a variety of manners, essentially anything that is non-linear in the system can cause intermodulations (transformers, transistors, pentodes, poor layout, poor topology choices, etc.)- I have seen a dirty switch make a big difference. One inharmonic intermodulation point occurs at the feedback node of any amplifier (Crowhurst), which is one reason I avoid using loop feedback. Interestingly, the ear has the ability to hear about 20 db into a noise floor of the right type (an exception to the masking rule, apparently) if that noise floor is similar to the sound of wind (I personally suspect this is a consequence of survival/evolutionary influences). If the noise floor is composed of harmonics, intermodulations and inharmonic artifacts, the ear cannot penetrate that floor and it will thus describe the lowest level of detail available. Zero feedback circuits can have a more towards a natural noise floor similar to wind; this may explain why they tend to have more detail even when the apparent noise floors measure at the same level. More research is needed IMO.

Was that a yes?

Tim

 

[Ethan Winer]

Do you disagree with the position that electrical performance is the sole feature of cable audibility in a blind test?

This nails it. If someone doesn't understand the relation between electrical performance and audibility, or even that there is such a relation, I'm not sure what could be said to convince them.

--Ethan