[AutumnCrown]

People often speak of fatigue as something that is cumulative, that it takes minutes or hours for a strident headphone to become fatiguing. This suggests that there is a complex psychological time interaction between the equipment and the brain. The time limitations of blind tests have been noted by others. In plain English, because time magnifies certain aspects of sound, there may be effects too subtle to be noticed in time-limited blind tests, and the fatigue effect is evidence of this.

Does this make any sense? I've had a lot of coffee today, so it might not 

 

 

[castleofargh]

a blind test(or any test) must first have a specific purpose. usually it's simply to notice if 2 devices, files, or situations, can be audibly different to the subject being tested. we rarely suggest blind testing for more than that.
if your specific purpose is to know if a headphone is fatiguing, you will first need to find a proper testing method that could reveal this. for something like fatigue, it might be very tricky.

 

[AutumnCrown]

Yes I agree that goal of a blind test should be specific and framed as such, and it should not be thought of as proof of devices' absolute quality, though it seems that that is what it is most often used for. My observation is that this is one noticeable time dependent listening effect, and my intuition is that it is probably the tip of the iceberg, which means, as you say, that blind-testing should not be the ultimate quality test of a piece of a equipment.

 

[spruce music]

You can use tools developed by the food industry for tastes and food satisfaction.
 
They are still double blind.  Though not usually ABX tests.
 
One good for testing a particular perception is called a duo-trio test.
 
You would play a reference sound or track.  Then you have two other tracks.  You have one which is a copy of the reference, and one which is different (more fatiguing or more bass or more treble or more spacious whatever you wish to test for). 
 
So for instance you would play the reference then ask the testee to listen to the other two and pick which has more bass just for example.  You need the file to really have more bass.  Then you can determine if they can reliably spot the amount of extra bass or if they are getting results equivalent to guessing. 
 
So doing this for fatigue you need to know what is causing fatigue and be able to put variable amounts of it in a test file.  If you can do that, you can test for how much is perceived and at what point it is a non-issue.

 

[upstateguy]

Isn't it easier to run a null test like DiffMaker to see if there are any differences?

 

[AutumnCrown]

  You can use tools developed by the food industry for tastes and food satisfaction.
 
They are still double blind.  Though not usually ABX tests.
 
One good for testing a particular perception is called a duo-trio test.
 
You would play a reference sound or track.  Then you have two other tracks.  You have one which is a copy of the reference, and one which is different (more fatiguing or more bass or more treble or more spacious whatever you wish to test for). 
 
So for instance you would play the reference then ask the testee to listen to the other two and pick which has more bass just for example.  You need the file to really have more bass.  Then you can determine if they can reliably spot the amount of extra bass or if they are getting results equivalent to guessing. 
 
So doing this for fatigue you need to know what is causing fatigue and be able to put variable amounts of it in a test file.  If you can do that, you can test for how much is perceived and at what point it is a non-issue.

What I'm saying is that certain differences that people notice appear to be cumulative (fatigue from strident treble being the most common), so subtle ones are only noticeable with lengthy trials. I'm not sure if you are saying that this method would circumvent the need for long trials? I can't think of any way to simulate this effect, because it appears to operate on a more subtle level than conscious experience.

 

[Ruben123]

Yes I agree that goal of a blind test should be specific and framed as such, and it should not be thought of as proof of devices' absolute quality, though it seems that that is what it is most often used for. My observation is that this is one noticeable time dependent listening effect, and my intuition is that it is probably the tip of the iceberg, which means, as you say, that blind-testing should not be the ultimate quality test of a piece of a equipment.

You can take as long as you want going an abx test. What's the difference then with a normal signed test, other than that you can't see?

 

[AutumnCrown]

You can take as long as you want, but few testers spend many hours in an abx test at all, let alone spend hours listening to stimulus a before switching to stimulus b. The nature of cumulative effects may mean that at the moment of switching stimuli they would not notice any difference, and only after a delay begin to be noticeably affected by the difference. Over long periods, normal metabolic rhythms, mood shifts, boredom, etc may cloud the picture for the subject, even if the effect is "real". Like the difference between drinking 4% and 5% ABV beer, the difference may be to small to measure using psychological tests, but still significant over the long term. If you know of any hours-long abx tests I'd be obliged if you would share them.

 

[castleofargh]

  You can take as long as you want, but few testers spend many hours in an abx test at all, let alone spend hours listening to stimulus a before switching to stimulus b. The nature of cumulative effects may mean that at the moment of switching stimuli they would not notice any difference, and only after a delay begin to be noticeably affected by the difference. Over long periods, normal metabolic rhythms, mood shifts, boredom, etc may cloud the picture for the subject, even if the effect is "real". Like the difference between drinking 4% and 5% ABV beer, the difference may be to small to measure using psychological tests, but still significant over the long term. If you know of any hours-long abx tests I'd be obliged if you would share them.

but again few people do abx for anything but see if they can perceive an audible difference. if I go and measure frequency response, you can't just take the result and tell me there is jitter. my test was specific to frequency response and not intended to know anything else. a blind test is the same, depending on what is being tested, a different protocol may be required. if the guy was trying to notice an audible difference, he did or didn't. that says nothing about fatigue or what he took for breakfast. not because the test is wrong, but because it wasn't the purpose of that one test.

 

[cjl]

Is there any proof of this "fatiguing" character not being audible in a fast-switch a/b? I'd definitely believe that you might not be able to tell which one would be the fatiguing one, but I would think that it would at least be audible that there was a difference.

 

[spruce music]

  Is there any proof of this "fatiguing" character not being audible in a fast-switch a/b? I'd definitely believe that you might not be able to tell which one would be the fatiguing one, but I would think that it would at least be audible that there was a difference.

In my experience a difference that causes fatigue, while not heard as such is going to be easily ABX'd in short term comparisons as different. 

 

[castleofargh]

most likely.
maybe with the exception of high amplitude ultrasounds, that could still ruin the ear without being readily noticeable. but I suspect that would lead to a bunch of other problems on the sound system. and those may still create an audible difference.

 

[AutumnCrown]

  Is there any proof of this "fatiguing" character not being audible in a fast-switch a/b? I'd definitely believe that you might not be able to tell which one would be the fatiguing one, but I would think that it would at least be audible that there was a difference.

The thrust of my argument is not about fatigue, but about the experience of listening over the longer term in general. If a subtle difference becomes a large difference with the addition of time, might a difference below the threshold of human abx testing and conscious awareness also become larger with the addition of time?

 

[castleofargh]

 

  Is there any proof of this "fatiguing" character not being audible in a fast-switch a/b? I'd definitely believe that you might not be able to tell which one would be the fatiguing one, but I would think that it would at least be audible that there was a difference.

The thrust of my argument is not about fatigue, but about the experience of listening over the longer term in general. If a subtle difference becomes a large difference with the addition of time, might a difference below the threshold of human abx testing and conscious awareness also become larger with the addition of time?

 
as you're using this to criticize blind test, there should be an implied second device or second sound used in the blind test. so we have 2 kinds of situations that could lead to one device being fatiguing or more fatiguing:
1/ all the situations where a blind test would let us notice a difference of any sort, and one sound is more fatiguing in the long run. that couldn't possibly account as a blind test flaw. the test was to find any audible difference and we did.
2/ cases of 2 things sounding close enough to make us fail a blind test, but still different enough to have a different lasting impact, based on sound alone!!!!!!!!!
 
 do you know of situations such as 2/ in real life? with failed blind tests that can be verified objectively to have audible impact in the long? the hypothetical example I gave above with ultrasounds of high amplitude is the only thing I can think of right now that might work. but I'm not sure it would and I don't know of a test that checked if it could. + that shouldn't happen often between the usual content of music and the low pass filters used in a DAC.
 we can't possibly blame blind tests for something that we think may perhaps exist. first find out if it can exist, then blame blind tests for missing it.

 

 

[AutumnCrown]

Perhaps the title of this thread is misleading. The context in which I see blind tests used, as I've stated, is as be-all end-all tests of fidelity and quality. Hence why objectivists use them frequently to debunk subjectivists' claims. I'm not criticizing blind tests that are presented, correctly, as limited and context-dependent tests. But assuming it does actually exist, how could I demonstrate a 2/ in real life? I'm suggesting that the tools we have for measuring the way people experience qualities of sound fail to capture subtle time-dependent qualities. If it exists, and is for the time being unprovable, is it not still worth thinking about? Much of modern physics operates on a similar premise.