If this are differencies in sound, than you have to use (or need to have) more sensitive ears
, not "to hear" using eyes and ilusions.. This will be scientificaly appropriate view.Understood from your position. If you were here you would have a different take because you would see details missing in your present information set. No matter, you can trust my judgment and know how or not. If not, then we are stuck because we need money if we are to run proper research.
Think I have said enough, we are back into the infinite argument territory. No point in going on. Best to stop here since I know I won't last until infinity time.
If you like to check the numbers for "30% of the people on here" (even if strechted far out in using all the members) and the for "5%" of the population it really could be true ......
Poor analogy, because we can often build a better detector. We can't build better ears.Markw4 said:
ABX would confirm that a claim of 'obvious' is actually true. There are lots of false claims of 'obvious'. ABX might even show us how obvious. It is reasonable to suppose that some less obvious difference would only be reliably heard by some people, so their score would stay up while others' scores would go towards random guessing.
"Well known" on what basis? That it gets the 'wrong' result according to some people?
Then do so. Nobody is wedded to ABX. Find another test which allows hearing to be the arbiter of sound. No peeking alowed.
Yes, 5% of who, those who's cats expressed a preference? (from their food bowl in the kitchen)If you like to check the numbers for "30% of the people on here" (even if strechted far out in using all the members) and the for "5%" of the population it really could be true ......
Or, of course, you are hearing a difference because you expect to - not necessarily the same as "sighted listening".
What's the methodology that "removes sighted listening effect"?
One of the interesting results from the Richard Clark Amplifier Challenge was the tests we did to determine bias: We had the listeners do the same tests twice, once writing their observations, and the next battery while encouraged to discuss the results as they happened. The tests done in written form resulted in fully random correlation to either the actual source or between the listeners. The results while discussing ended up correlating highly between listeners, but once again a random correlation to the actual source.
Likewise our tests using a light to identify the source (but which actually was not) resulted in high correlation of differences to the light, not to the actual source.
Listening to the test subjects describe what they heard convinced me people actually "heard" differences. That is, their auditory cortex processed the sound in a way that made it sound different, but the factor causing the differential processing was not cochlear output, indeed it could be visual cortex input, or just plain expectation bias. You can imagine you are drinking milk while sipping orange juice...for a short while.
Doing these tests was fun for a while..and then (for me) it was not. I won't belabor the arguments that broke out when people realized they were unable to reliably identify sources...I have no interest in that aspect of testing, and egos are apparently difficult to control...I was able to identify the less egotistical engineers by their smile and lack of ego when discussing the tests. Several made suggestions which helped us design other test protocols. After all it was just for fun and learning, no human was hurt during the tests.
Have a great weekend, all!
Howie
Attachments
Of course not. I expected to hear no change for a small adjustment. That's the problem with trying to describe events here in text. People look for any explanation their imaginations can construct and believe them as though they are fact. Another demonstration System 2 is lazy.
To provide a little more detail after my surprise at finding a small adjustment made an audible difference, I adjusted for a measured distortion null of the H3. Didn't like the sound, presumably because of remaining H5 and H7 that could not be changed. Decided to adjust H3 for my preferred sound. Null was at -157 IIRC, preference was at -150. Did a repeated search on each side of the preferred setting much like tuning in IF transformer slug with a wide-ish peak (although this was not that wide). Look at the sides of the curve and find the center. Converged on -150 multiple times homing in on my preferred sound, left it there. That was the most reliable fit to my preference I could find. There are some more details to this I won't go into to keep it brief, just stuff having to do with tuning sensitivity, we can skip it for now.
My son walks in from outside. Last time he was there I was swapping opamps which we both heard the same way. He is getting much better at hearing distortion with practice. I said, "listen to both speakers," and nothing more. It was a slight misdirection as last time I had them sounding different. He walked up and listened to the speakers separately and together. He frowned and said they sound the same as each other, but they are less bright and less distorted this time. He was exactly right according to the measured adjustments I made.
From previous measurements I also know H2 was sitting at -120dB +- 1dB-2dB when I started. The error band is because of noise filter settling time. I don't want to wait forever for the noise to settle.
We get these things right all the time with at least one of us blind, sometimes both. Occasionally there are differences too small to call. Mostly they are sufficient to discern blind, at least if ear fatigue has not set in, which usually we don't let that happen.
Now, I am sure people can dream up all kinds of stories about how there was this or that bias or mistake, blah, blah, blah. Have fun with your imaginations.
BTW, in the case of the H3 adjustment, I think most people here could hear that if they were asked to listen for it. They might need to have a short training session switching back and forth to learn what to listen for, but in this case it was not hard.
Okay, I really want to go back offline on this. Please stop talking to me and I will try to return the favor.
Last edited:
True.
That we are dealing with audio and not with biological reproduction is a sure sign of …
George
I understand, and share the interest, but nevertheless thought that it doesn´t help only to criticize one side while the other is acting quite similar.
Therefore i metioned that for example in case of a positive "ABX result" the "its just fun" part usually suddenly ends.
Imo the test results mmerrill99 linked to are a typical examplefor this mechanism. (Btw this mechanism was one of the reasons why i started the thread about the evidence that people would accept in case of results contradicting their prior beliefs)
Wrt the meta-discussion; i would include the people discussion the habits of others too; i might be mistaken but the word "excuse" is/was imo mostly used with the negative connotation of "unjustified face-saver" which seem to be unfair as we know that there exist good explanations for negative results although listeners are principally able to differentiate between two stimuli.
Being sceptical is imo good/justified but it should be balanced.....
Last edited:
Thanks for the detailed and informative reply.
My only small point - the "system 1, 2" stuff is a very early hypothesis only. Some find is useful - amongst folk I know (including research psychologists and AI researchers) it's considered somewhere between a useful tool and not at all relevant. We know so little really; this comment is not intended as a criticism of you!
Otoh good analogy, because as a physicist you´d never use detectors without checking their properties but that´s what is routinely done in these "it´s just fun tests" ; unfortunately often in less "fun based" approaches as well.
Isn´t that first of all a matter of semantics or synchronization before drawing any further conclusions?
Let me reintroduce the "gorilla in our midst" example. Is that in the compartment of "obvious" ? I´d say so, but obviously it can remain undetected if only the distraction is sufficient.
Another example would be a difference first missed in a controlled listening test, but after being pointed to it is "heard" always under the same test conditions.
Does this difference then qualify for the descriptor "obvious" ?
Obviously if something is an "obvious" difference depends on a lot of variables and one shouldn´t pretend that there exists broad agreement about such terms / their meaning.
More precisely you/me/us as third party observers simply don´t know what is/was tested, stating the above as fact is misleading as _we_ _don´t_ _know_ .
We only know that generally such tests can be associated with a high risk of bias impact, but if a specific test was prone to that bias, nobody knows without examining it.
If we are trying to provide corrobation acceptable for third partys we normally don´t use "sighted listening" tests as one can not show internal validity.
Doing something "blind" does not ensure internal validity either; so it doesn´t help in comparison to "sighted listening tests" as one still can´t show internal validity. What is difficult about this?
Is that the guy writing, who takes often all literally? Can i have that guy back, please?
Come on, as a scientist, weren´t you getting a bit suspicious if the data fits in a way that was too good to be true?
You evaded the point that mixing results from different experiments over all possible effects done under virtually no quality control together to declare "indistinguishability" can´t be reflecting the "true spirit" of good experimental design and analysis.
And i´m indeed quite confident that physicists would have never done much progress if they´d follow such incredible sloppiness.
And if you only could you would deny that the "gorilla" really was there but unfortunately it was......
I totally agree that people should restrain from these "night and day " descriptors, but if you insist on "take it literally" please follow it in every direction not only when targeting "audiphiles".
Which is a clear bias and you should be aware of it.
- Status
- Not open for further replies.