Yes and no. In reality a listening test is a challenge for the listener too. Although it isn´t test for the person it is for the listening abilities and given the fact that often lengthy discussions preceed the test where sanity and physical laws were used as arguments i think we can assume that the person will feal challenge quite often.
If we want to know if our test is good, we have to use positive controls (a positive control is a difference that has to be detected under test conditions) i think it is obviously a "test the persons ability" too.
"We" know that several members have tried to get ABX result with the files under test, and also with the original file posted later. Based both on posts here and personal messages that I have received. The results of these trials were negative.
PMA, of the people who you know tried ABX, how many reported results indicating negative correlation?
Of course, but i´d assume the number of people trying and reporting about that was even lower than the number of people doing the poll.
So, if the number was to small in the poll to allow....... , it seems strange that the lower number would allow an even more advanced conclusion (see the generalized messages).
There's no advanced conclusion at all, its very simple: No one has reliably demonstrated to hear a difference between these files, up till now.
For people who believe reliable=ABX, that's probably true.
For people with different beliefs, other interpretations are possible.
For people with different beliefs, other interpretations are possible.
I'm open to all reasonable interpretations.
Up till now we have one person who was able to identify the opamps through mail to PMA. This is very easy to cheat with free software available to everyone and some spec sheets.
And there's the anecdotal evidence in the form of: File X sounds ....... compared to file Y.
So sceptical as a sane person should be, this is as acceptable as me saying I took a dump on the moon now give me the Google lunar x prise. Not going to happen.
What I want is someone saying, this specific thing is the "audible tell" witch I use to discriminate these files and I did not cheat because here's the proof. Now please other people, try to duplicate my results. Doesn't have to be ABX, there's lots more protocols that can be used. But all of these protocols should only use the ears as input to our brain when trying to discriminate these files.
"Up till now we have one person who was able to identify the opamps through mail to PMA"
I think I missed that post?
I think I missed that post?
http://www.diyaudio.com/forums/ever...6-24-listening-test-opamps-3.html#post5111330
So no identification of the opamps. (I can't change that in my post.)
But, I assume, noise levels have been used.
Last edited:
I was the person PMA was referring to. At that time PMA told me no on else had done that well, so I posted in considerable detail what I did and told other people how they could try the same thing. I don't know if anybody did. That was this post: http://www.diyaudio.com/forums/ever...6-24-listening-test-opamps-5.html#post5111943
Later after the test ended, PMA reported the results as he saw them. I took a different view of the results and posted my thoughts here: http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-24.html#post5118416
Also, I discussed trying to use Foobar ABX testing, which resulted in the discovery that I wasn't the only one with a reverse correlation tendency. PMA also had the same effect, but thought it meant he was guessing. http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-17.html#post5116915
jcx later commented to the effect that reverse correlation meant something was being heard.
There is also a lot of other interesting posts. I think some people were able to hear the differences better than I did, but they described them differently and in a way possibly a little harder for some people to correlate with op-amp measured performance.
My conclusion was not that we proved anything definitively, but that a least the results suggest hearing such small differences is likely to be real to some extent or other, and that ABX may not be the best test for measuring that, although it does happen to be the hardest test to cheat on that we have available today for free. That doesn't mean other types of tests can't be developed that are just as hard or harder to cheat on.
Finally, I would vote for more more and better research, something that could be published and that would very likely be a significant update to the existing literature.
Later after the test ended, PMA reported the results as he saw them. I took a different view of the results and posted my thoughts here: http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-24.html#post5118416
Also, I discussed trying to use Foobar ABX testing, which resulted in the discovery that I wasn't the only one with a reverse correlation tendency. PMA also had the same effect, but thought it meant he was guessing. http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-17.html#post5116915
jcx later commented to the effect that reverse correlation meant something was being heard.
There is also a lot of other interesting posts. I think some people were able to hear the differences better than I did, but they described them differently and in a way possibly a little harder for some people to correlate with op-amp measured performance.
My conclusion was not that we proved anything definitively, but that a least the results suggest hearing such small differences is likely to be real to some extent or other, and that ABX may not be the best test for measuring that, although it does happen to be the hardest test to cheat on that we have available today for free. That doesn't mean other types of tests can't be developed that are just as hard or harder to cheat on.
Finally, I would vote for more more and better research, something that could be published and that would very likely be a significant update to the existing literature.
Last edited:
Imagination and illusion made you guess and by accident you guessed an order that without 2134 could make sense. However the test and your order included 2134, to reveal the amount of guessing. You disagree, because you also only get guessing scores for abx, and we all should know, a double negative result makes a positive.
Because 3 times left equals right once too, you argue the foobar abx guessingscore is incorrect.
Because 3 times left equals right once too, you argue the foobar abx guessingscore is incorrect.
Do you, or any one else, have any peer reviewed articles on this?
I really hope there's some interesting research on this, but I don't know about it, until some research shows up, I'll put this in the realm of wishful thinking.
That people described the differences differently, is a strong clue that they were imaging things.
There are lots of other test protocols out there, ABC/HR, triangle, MUSHRA to name but a few.
Ignoring the vast amount of perceptual testing already done in this field. Looks to me like your trying to find excuses.
Cognitive dissonance is hard to overcome, been there done that, green ink still on many CD's, took me years but I've managed. As have others see Billshurv post: http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-58.html#post5129569
I'm sorry to point this out, but its just to funny:
http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-27.html#post5118686
ww being the source file.
http://www.diyaudio.com/forums/ever...-24-listening-test-opamps-27.html#post5118686
ww being the source file.
Actually, you are much closer than you think. Imagination and illusion made you guess at what I did. However, your guess is wrong, but that's okay. Better luck next time.
I think the negative/reverse correlation being attributed to consistently getting the abx flipped.
E.g. some time ago I tried to differentiate between files with different white noise bandwidths (e.g. same white noise file rolled off with different low-pass filters, not a small effect). I think I scored something > 16/20 on the 16 kHz lowpass vs the 22.05 kHz source on an ABX -- except I that scored 16/20 making the *wrong* guess. Whether that still falls in guessing is a legit question since 20 trials isn't too many, but points to me hearing the difference even if I screwed up the attribution portion (consistently!)
Was still rather humbling, to say the least.
E.g. some time ago I tried to differentiate between files with different white noise bandwidths (e.g. same white noise file rolled off with different low-pass filters, not a small effect). I think I scored something > 16/20 on the 16 kHz lowpass vs the 22.05 kHz source on an ABX -- except I that scored 16/20 making the *wrong* guess. Whether that still falls in guessing is a legit question since 20 trials isn't too many, but points to me hearing the difference even if I screwed up the attribution portion (consistently!)
Was still rather humbling, to say the least.
It is you who makes himself funny.
Your fixation, your pretentious presumption that the source file sould be sacred, that a source file by definition is perfection itself, that the source file Is The One That sounds Good --- now that is funny.
In fact, the trick, the twist in this test was this 'secondary' feature: that a distorted, compressed, but 'supposed to be Hi-Res' source file had simply reversed the partecipants preferences: they voted for the samples the farthest away from the original.
Because, the other trick in the package was that they were not given, they got robbed of the possibility to listen to the original. So, apart from the experimenter, loaded with his biases and dirty tricks, nobody new how the original should sound.
Do you really dare to claim that the result would have been this same outcome, given the partecipants the chance to listen to the original?
(and register how strange sounding it is?)
Sorry for You,
George
- Status
- Not open for further replies.