Audiophile Placebo

[Ashirgo]

The only thing that I do not doubt here is that audiophiles' perception should be tested if their hearing really exceeds human/normal capabilities. You can say that they learned it or improved their perception, if they did, it should be reflected in a measurement.

 

[b0dhi]

Quote:

Originally Posted by Ashirgo /img/forum/go_quote.gif The only thing that I do not doubt here is that audiophiles' perception should be tested if their hearing really exceeds human/normal capabilities. You can say that they learned it or improved their perception, if they did, it should be reflected in a measurement.

A good start to learning about human learning when it comes to audio analysis is here: Human echolocation - Wikipedia, the free encyclopedia. You'll note that these people are not audiophiles.

Measurements in audio are primitive when compared to most other fields of engineering and science. In particular, the dynamic performance of the audio chain is poorly gauged IMO. The measurements we usually make can be very accurate, but thorough they are not. On the other hand, the ears may not have the same precision or accuracy, but they have a tremendously powerful signal analysis tool working on the audio data - the brain. Not only is it powerful, but it can learn.

So, simply saying "if you can hear it, prove it" is not helpful because things aren't so black and white. Our mental "music analysis" software did not evolve or learn to make conscious comparisons in an ABX. Assuming that the best we can do in an ABX represents the best we can hear is, therefore, a false assumption. That's not to say it's useless, but the conclusions we draw should be tempered by reason. Hirsch explained this much better than I can earlier in this thread, so I don't know why the discussion is going around in circles.

 

[ILikeMusic]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif So, simply saying "if you can hear it, prove it" is not helpful because things aren't so black and white. Our mental "music analysis" software did not evolve or learn to make conscious comparisons in an ABX. Assuming that the best we can do in an ABX represents the best we can hear is, therefore, a false assumption. That's not to say it's useless, but the conclusions we draw should be tempered by reason. Hirsch explained this much better than I can earlier in this thread, so I don't know why the discussion is going around in circles.

Because that argument is little more than a subterfuge to avoid the simple reality that people are often unable to demonstrate what they claim.

 

[nick_charles]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif Measurements in audio are primitive when compared to most other fields of engineering and science.

Really ? - we can measure level differences down to 10000ths of a db routinely, jitter down to 1,000,000,000,000ths of second, distortion down to 10,000ths of a percent, same with speed deviations and noise. Perhaps crude in the space science sense but surely sensitive enough ?


Quote:

In particular, the dynamic performance of the audio chain is poorly gauged IMO.

See above, even with my $30 ADC I can get a differentiation between stimuli at a level way better than any human alive, imagine what you can do with a $50K test set-up.

Quote:

The measurements we usually make can be very accurate, but thorough they are not

.

So what is missing ? What parameters (real ones) are missed ? . Please don't say PRaT , this is the science forum

Quote:

On the other hand, the ears may not have the same precision or accuracy, but they have a tremendously powerful signal analysis tool working on the audio data - the brain. Not only is it powerful, but it can learn.

People learn , nobody doubts that , train listeners before you test them , train them for a 1000 hours if you like, you will still hit the limits of human perception , well known limits, and even if yoiu raise those limits by an order of magnitude....

Quote:

So, simply saying "if you can hear it, prove it" is not helpful because things aren't so black and white. Our mental "music analysis" software did not evolve or learn to make conscious comparisons in an ABX.

As mentioned above people learn , they can learn to do better in ABX tests, sighted test are wholly unreliable by any rational criteria.

Quote:

Assuming that the best we can do in an ABX represents the best we can hear is, therefore, a false assumption.

See above

 

[b0dhi]

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif Really ? - we can measure level differences down to 10000ths of a db routinely, jitter down to 1,000,000,000,000ths of second, distortion down to 10,000ths of a percent, same with speed deviations and noise. Perhaps crude in the space science sense but surely sensitive enough ? See above, even with my $30 ADC I can get a differentiation between stimuli at a level way better than any human alive, imagine what you can do with a $50K test set-up.

I said "The measurements we usually make can be very accurate, but thorough they are not. On the other hand, the ears may not have the same precision or accuracy...". I am talking about analysis, you are talking about measurement.


Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif So what is missing ? What parameters (real ones) are missed ? . Please don't say PRaT , this is the science forum People learn , nobody doubts that , train listeners before you test them , train them for a 1000 hours if you like, you will still hit the limits of human perception , well known limits, and even if yoiu raise those limits by an order of magnitude....

I'm not even sure where to begin unraveling this.

First -
"Train them for 1000 hours"?

We are trying to determine maximum hearing capability of the ear/brain system. The ABX test, though, can only validly infer the maximum performance of the listener in performing the training excercise.

Further, the very act of requiring a self-reported conscious comparison by the listener as the test mechanism limits the domain of validity of the test. Conscious perception is very small relative to total (i.e., subconscious inclusive) perception. For example, autistic people are sometimes able to remember each note accurately in an entire symphony in one listening, or remember every tiny detail in a complex image with a quick glance. Since listening to music is not a process whereby one sits down and consciously compares sound, a test depending on self-reported conscious comparisons may not expose subtle changes in the listener's mental/brain state which lie beyond the perception of the listener themself. Objective measurements are required.


Secondly-
"train them for a 1000 hours if you like, you will still hit the limits of human perception"

No, actually, you won't. Even if you trained them in every possible hearing-related function, for 100000 hours, you wouldn't hit the limits of human perception - you'd hit the limits of conscious human perception. They can't be considered the same thing at all, and the experiment changes entirely when dealing with one instead of the other.

Moreover, the test you suggest is itself flawed since you're measuring two variables - the effectiveness of the training and the performance of the listener. If the test fails, which do you blame? Also, in reality the test would also most likely very soon hit the limits of training effectiveness, human fatigue or the budget.


Thirdly,
"you will still hit the limits of human perception , well known limits"

We know for example that the ear is sensitive to displacement of air of about 2nM. That determination was made with objective measurements of the ear itself, not subjective ABX, so it doesn't make any difference how well the listeners are trained - the limit will still apply equally to the untrained as it does to the best trained. Other of the "known" limits are, however, determined by self-reporting. Those types of determinations have a domain within which they're valid (conscious comparisons, if that be the case), and beyond that domain have not been shown to be valid.

 

[nick_charles]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif I said "The measurements we usually make can be very accurate, but thorough they are not. On the other hand, the ears may not have the same precision or accuracy...". I am talking about analysis, you are talking about measurement.

Agreed but the two are not unrelated an analysis without some form of obective measurement is incomplete.


Quote:

We are trying to determine maximum hearing capability of the ear/brain system. The ABX test, though, can only validly infer the maximum performance of the listener in performing the training excercise.

Once you have trained subjects to improve their discrimination ablities you test them with the real task, you train them until they consistently show a lack of improvement.

Quote:

Further, the very act of requiring a self-reported conscious comparison by the listener as the test mechanism limits the domain of validity of the test.

No verbal protocol is required at all. They have a rig and a choice of two buttons , no explicit verbalization is required. I have used verbal protocols in some of my research and they are useful in some settings , they are not required here.

Quote:

Conscious perception is very small relative to total (i.e., subconscious inclusive) perception. For example, autistic people are sometimes able to remember each note accurately in an entire symphony in one listening, or remember every tiny detail in a complex image with a quick glance. Since listening to music is not a process whereby one sits down and consciously compares sound, a test depending on self-reported conscious comparisons may not expose subtle changes in the listener's mental/brain state which lie beyond the perception of the listener themself.

What is going on in the brain of the listener is just irrelevant, we cannot see it and speculating about it in this context is not helpful. Can they tell a difference or not, simple. If they cannot what evidence do we have that any "subtle" differences between A and B are perceived and not imagined

Quote:

No, actually, you won't. Even if you trained them in every possible hearing-related function, for 100000 hours, you wouldn't hit the limits of human perception - you'd hit the limits of conscious human perception.

Unconscious perception is meaningless. Sometimes you can detect a difference without being able to explicitly describe it , but this does not mean that some magical unconscious process is at work merely that the pattern recognition capabilities do not lend themselves to verbalization, but as I have said this is not required for ABX tests

Quote:

They can't be considered the same thing at all, and the experiment changes entirely when dealing with one instead of the other.

Ever heard of transferability ?

Quote:

Moreover, the test you suggest is itself flawed since you're measuring two variables - the effectiveness of the training and the performance of the listener.

No, the training is merely a prelude to the test. It does not matter how good a subject is at the beginning or how much or little they improve. The point is to get them to be as good as *they* can be. So to make the test as fair as possible.

Quote:

If the test fails, which do you blame? Also, in reality the test would also most likely very soon hit the limits of training effectiveness, human fatigue or the budget.

The test neither fails or succeeds the test shows whether a subject can detect a difference or not. As for training effectiveness, well you have breaks to deal with fatigue and you spread the testing out over a period of time.

Quote:

Those types of determinations have a domain within which they're valid (conscious comparisons, if that be the case), and beyond that domain have not been shown to be valid

Unconscious comparisons make no sense at all.

 

[b0dhi]

You seem to be assuming I'm positing subjective/unblinded tests. I'm not. As I said in my previous post, I think objective tests are needed. An ABX is not an objective test, it's a self-reported and therefore subjective test albeit with objective interpretation of the data.

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif No verbal protocol is required at all. They have a rig and a choice of two buttons , no explicit verbalization is required. I have used verbal protocols in some of my research and they are useful in some settings , they are not required here.

Where did verbosity come into it? I didn't mention any such thing. Whether the test is administered by the press of a button or by word, it still is a self-reported subjective test by the listener.

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif What is going on in the brain of the listener is just irrelevant, we cannot see it and speculating about it in this context is not helpful. Can they tell a difference or not, simple. If they cannot what evidence do we have that any "subtle" differences between A and B are perceived and not imagined

This is where we diverge. As I said, I'm discussing ways to determine the maximum hearing capability of the human ear/brain system, with relation to how it affects listening to music.

What goes on in the mind/brain of the listener is eminently important for making that determination because there's no guarantee that conscious self-reported comparisons by the listener represent:

A. all of the as yet untrained sound analysis capabilities of the brain

B. all the specific types of sound analysis the brain deems significant when it comes to experiencing music, even if the listener is consciously aware of them.

C. all actual audio differentiations performed by the brain - which affect its state - but which are not perceived as consciously observed differences to the listener. These are not testable in an ABX but are relevant when listening to music.


Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif No, the training is merely a prelude to the test. It does not matter how good a subject is at the beginning or how much or little they improve. The point is to get them to be as good as *they* can be. So to make the test as fair as possible.

I agree with most of this, but it missed the point. Clearly the training needs to get them to the point where they're "as good as they can be" for the ABX to represent their maximum hearing capability. But when will we know they're at their best as opposed to when the training is ineffective, or has reached its peak effectiveness? In both cases the tests will show gradual performance plateuing.

So, training is not just a "prelude" - the test has a strong dependence on it. The test is also measuring the training, not just the listener. A flaw in the training can't be distinguished with hitting a limitation in hearing (without the training itself being analysed).

Also, the idea of training someone for maximum performance in using a perceptive sense is quite absurd when it comes to what we're discussing IMO. It's like saying "we'll train them to use their eyes to their best ability". It's a nonsense statement because there's no measure of overall "wellness" in using one's eyes - there are only specific functions and behaviours that can be learned and tested specifically. You can test the resolution of vision in the periphery, the resolution in the center, the closest focussing distance of the eyes, etc, but if those are all you train and test for, the tests won't tell you much about, for example, whether the viewer experiences Change Blindness, how wide their field of attention is, what the quickest observable sudden change in visual stimulus is, or any number of other factors that the "wellness" training doesn't encompass.

 

[nick_charles]

Quote:

Originally Posted by b0dhi /img/forum/go_quote.gif You seem to be assuming I'm positing subjective/unblinded tests. I'm not. As I said in my previous post, I think objective tests are needed. An ABX is not an objective test, it's a self-reported and therefore subjective test albeit with objective interpretation of the data. Where did verbosity come into it? I didn't mention any such thing. Whether the test is administered by the press of a button or by word, it still is a self-reported subjective test by the listener. This is where we diverge. As I said, I'm discussing ways to determine the maximum hearing capability of the human ear/brain system, with relation to how it affects listening to music. What goes on in the mind/brain of the listener is eminently important for making that determination because there's no guarantee that conscious self-reported comparisons by the listener represent: A. all of the as yet untrained sound analysis capabilities of the brain B. all the specific types of sound analysis the brain deems significant when it comes to experiencing music, even if the listener is consciously aware of them. C. all actual audio differentiations performed by the brain - which affect its state - but which are not perceived as consciously observed differences to the listener. These are not testable in an ABX but are relevant when listening to music. I agree with most of this, but it missed the point. Clearly the training needs to get them to the point where they're "as good as they can be" for the ABX to represent their maximum hearing capability. But when will we know they're at their best as opposed to when the training is ineffective, or has reached its peak effectiveness? In both cases the tests will show gradual performance plateuing. So, training is not just a "prelude" - the test has a strong dependence on it. The test is also measuring the training, not just the listener. A flaw in the training can't be distinguished with hitting a limitation in hearing (without the training itself being analysed). Also, the idea of training someone for maximum performance in using a perceptive sense is quite absurd when it comes to what we're discussing IMO. It's like saying "we'll train them to use their eyes to their best ability". It's a nonsense statement because there's no measure of overall "wellness" in using one's eyes - there are only specific functions and behaviours that can be learned and tested specifically. You can test the resolution of vision in the periphery, the resolution in the center, the closest focussing distance of the eyes, etc, but if those are all you train and test for, the tests won't tell you much about, for example, whether the viewer experiences Change Blindness, how wide their field of attention is, what the quickest observable sudden change in visual stimulus is, or any number of other factors that the "wellness" training doesn't encompass.

You make a lot of interesting points, pretty much all of which I disagree with on pragmatic or definitional grounds , but I do not want to hijack this thread any further so I will answer by PM.

 

[IPodPJ]

I believe that scientific measurements in audio will never reveal why a human can hear that which a computer cannot, or rather reveal why we can perceive audio through tonal accuracy, emotion or otherwise. It's extremely similar to the reason websites ask you to type in a security code from the jumbled letters and numbers. Computers are not good when it comes to pattern extrapolation, but humans are. That is where the human brain excels. Likewise, our brains possess the ability to recognize patterns and signals in the music that enable us to hear cues for soundstage, imaging, detail, etc., but most of all our brains allow us to compare the audio to what we would hear in the natural world, something a computer cannot or would not understand.

More basic factors such as the electron have yet to be fully understood. Sure, we know that current flows through conductive elements, but we don't know why (in many instances) a silver cable produces a sound different to humans than a copper cable when the same measurements are taken for both. Quantum science is a field which is still in its infancy. Perhaps the LHC will reveal new insights into why an electron can be in several places at once, and perhaps why the subatomic universe is more connected with human perception and observation than we realize.

(Research quantum possibilities and collapsed wave functions. Also read an incredible article in this month's Discover magazine about the possibility of a Biocentric universe.) Our desire to hear better sound quality from our equipment could physically be affecting electron movement.

 

[nick_charles]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif but we don't know why (in many instances) a silver cable produces a sound different to humans than a copper cable when the same measurements are taken for both.

Since this is the Sound Science forum.

There is actually in fact no empirical evidence for this. I have seen not one verifiable case, and believe me I have looked long and hard, where anyone has shown any silver cable to be **reliably** audibly different from any copper cable.

I would be happy to look at any real evidence that anyone can point to but I haven't seen it. I exclude "My wife was in the kitchen" and "Clever Hans" type anecdotes and sighted tests.

 

[TheMarchingMule]

Quote:

Originally Posted by nick_charles /img/forum/go_quote.gif Since this is the Sound Science forum. There is actually in fact no empirical evidence for this. I have seen not one verifiable case, and believe me I have looked long and hard, where anyone has shown any silver cable to be **reliably** audibly different from any copper cable. I would be happy to look at any real evidence that anyone can point to but I haven't seen it. I exclude "My wife was in the kitchen" and "Clever Hans" type anecdotes and sighted tests.

To add a quick note onto this, doesn't anybody else find it interesting that the "personalities" of silver and copper's sound signature reflect their color? Silver is bright, and it's been stated that silver gives a clear, bright sound. Copper is a dull sheen, and its proposed sound signature is warm.

Hmmm...

 

[Real Man of Genius]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif I believe that scientific measurements in audio will never reveal why a human can hear that which a computer cannot, or rather reveal why we can perceive audio through tonal accuracy, emotion or otherwise. It's extremely similar to the reason websites ask you to type in a security code from the jumbled letters and numbers. Computers are not good when it comes to pattern extrapolation, but humans are. That is where the human brain excels. Likewise, our brains possess the ability to recognize patterns and signals in the music that enable us to hear cues for soundstage, imaging, detail, etc., but most of all our brains allow us to compare the audio to what we would hear in the natural world, something a computer cannot or would not understand. More basic factors such as the electron have yet to be fully understood. Sure, we know that current flows through conductive elements, but we don't know why (in many instances) a silver cable produces a sound different to humans than a copper cable when the same measurements are taken for both. Quantum science is a field which is still in its infancy. Perhaps the LHC will reveal new insights into why an electron can be in several places at once, and perhaps why the subatomic universe is more connected with human perception and observation than we realize. (Research quantum possibilities and collapsed wave functions. Also read an incredible article in this month's Discover magazine about the possibility of a Biocentric universe.) Our desire to hear better sound quality from our equipment could physically be affecting electron movement.

I think most in hear would agree that it is possible that science is as yet unable to detect XXXX. Where cables break down is when no one can detect the difference they claim to hear in a simple test.

 

[ILikeMusic]

Quote:

Originally Posted by TheMarchingMule /img/forum/go_quote.gif To add a quick note onto this, doesn't anybody else find it interesting that the "personalities" of silver and copper's sound signature reflect their color? Silver is bright, and it's been stated that silver gives a clear, bright sound. Copper is a dull sheen, and its proposed sound signature is warm.

It doesn't surprise me that a purely psychological effect would be based on color...

 

[IPodPJ]

Quote:

Originally Posted by TheMarchingMule /img/forum/go_quote.gif To add a quick note onto this, doesn't anybody else find it interesting that the "personalities" of silver and copper's sound signature reflect their color? Silver is bright, and it's been stated that silver gives a clear, bright sound. Copper is a dull sheen, and its proposed sound signature is warm. Hmmm...

Yes, I have also always had trouble with these "colorful" metaphors. I think those kinds of evaluations are definitely caused by placebo.

The differences I have heard between some copper and silver cables would be described as copper being more neutral and smooth sounding vs. silver which sounds slightly more detailed with a little less bass presence. But this isn't true for all, as there are so many different ways companies craft their interconnects.

So, if I see a review that uses the words bright and warm to describe silver and copper, I move straight on to the next review. Not to mention that cables and power cords are extremely hard to review anyway since they will do something different for every single piece of equipment.

 

[spanimal]

Quote:

Originally Posted by IPodPJ /img/forum/go_quote.gif I believe that scientific measurements in audio will never reveal why a human can hear that which a computer cannot, or rather reveal why we can perceive audio through tonal accuracy, emotion or otherwise. It's extremely similar to the reason websites ask you to type in a security code from the jumbled letters and numbers. Computers are not good when it comes to pattern extrapolation, but humans are. That is where the human brain excels. Likewise, our brains possess the ability to recognize patterns and signals in the music that enable us to hear cues for soundstage, imaging, detail, etc., but most of all our brains allow us to compare the audio to what we would hear in the natural world, something a computer cannot or would not understand. More basic factors such as the electron have yet to be fully understood. Sure, we know that current flows through conductive elements, but we don't know why (in many instances) a silver cable produces a sound different to humans than a copper cable when the same measurements are taken for both. Quantum science is a field which is still in its infancy. Perhaps the LHC will reveal new insights into why an electron can be in several places at once, and perhaps why the subatomic universe is more connected with human perception and observation than we realize. (Research quantum possibilities and collapsed wave functions. Also read an incredible article in this month's Discover magazine about the possibility of a Biocentric universe.) Our desire to hear better sound quality from our equipment could physically be affecting electron movement.

I bought the Belkin power filter on YOUR recommendations - for 300 aussie dollars it still is the single most valued investment I have made, bringing clearly audible changes to anything I connected to it. I still don't know why it works but it does, It taught me the term "noise floor" like the HD650 taught me the term "Audiophile" whilst the K701 showed me "detail" and "sound stage". I'm still not entirely certain of the term "pratt" hmmmm. Anyway, thanks mate.