Joshua,
Thank you for the undeserved flowers.
However much what I know that did not come from my rather limited experimentation, I learned from people like John, Charles, Nelson and Walt but also Hiroyasu-San, Nobu-San and Jean-San.
So I am very much the student to them and far from a master to join such a distinct group.
Ciao T
Thank you for the undeserved flowers.
However much what I know that did not come from my rather limited experimentation, I learned from people like John, Charles, Nelson and Walt but also Hiroyasu-San, Nobu-San and Jean-San.
So I am very much the student to them and far from a master to join such a distinct group.
Ciao T
Hi Thorsten,
Humility is a trait I appreciate in people. Actually, to my view, humility is a sign of true human greatness.
All Masters had teachers (save, perhaps, human beings like Buddha).
As far as I know myself, my mind and my pen (or keyboard) are matched, so, to my view, you definitely deserve what I wrote.
Humility is a trait I appreciate in people. Actually, to my view, humility is a sign of true human greatness.
All Masters had teachers (save, perhaps, human beings like Buddha).
As far as I know myself, my mind and my pen (or keyboard) are matched, so, to my view, you definitely deserve what I wrote.
There is a saying in my country: "Every day we learn something new and eventually die stupid".
μη δωτε το αγιον τοις κυσιν μηδε βαλητε τους μαργαριτας
υμων εμπροσθεν των χοιρων μηποτε καταπατησουσιν αυτους
εν τοις ποσιν αυτων και στραφεντες ρηξωσιν υμας
Hi,
I doubt it.
More likely that chinese proverb about learning how to fish, as opposed to being given a fish.
Ciao T
I doubt it.
More likely that chinese proverb about learning how to fish, as opposed to being given a fish.
Ciao T
Rather, fear of discreditation of design by implementing it's "improved" versions.
Hi,
Nope, rather:
"Try to teach a man how to fish and he will argue all day and night with you about the best lure, rod and reel instead of actually going and catching fish..."
Ciao T
....give a man a fishing rod and he will be drunk all day?
Nope, rather:
"Try to teach a man how to fish and he will argue all day and night with you about the best lure, rod and reel instead of actually going and catching fish..."
Ciao T
So if I go ahead and make an F5S (a "Shaman" version which deviates from Mr. Pass' amp design and which sounds awful... ), do I discredit Pass designs or do I just make a fool of myself?
Anyone with half a brain will be able to tell it's not the original design.
Those who don't even have that half of a brain will go about saying crazy things anyway (like get a "JC" marked PCB from eBay or come up with the Blowtorch schematic in their sleep and say this is the "John Curl sound").
So what?
@ SY & DF96
Although that seems to be funny or mysterious, in fact it is not.
It is well known from numerous experiments that even quite big differences can remain undetected if the human brain is dealing with confounding effects.
See for example the experiments of inattentional blindness, that i´ve linked a few times.
In addition it is confirmed that in controlled listening tests differences remained undetected although these differences were clearly audible.
So, if someone wants to investigate the audibility of differences the main thing is to design an experiment that only investigates the audibility of the difference, not other things the experimenter didn´t think about.
To quote from Bech&Zacharov:
Soren Bech, Nik Zacharov,
Perceptual Audio Evaluation–Theory, Method and Application
Although that seems to be funny or mysterious, in fact it is not.
It is well known from numerous experiments that even quite big differences can remain undetected if the human brain is dealing with confounding effects.
See for example the experiments of inattentional blindness, that i´ve linked a few times.
In addition it is confirmed that in controlled listening tests differences remained undetected although these differences were clearly audible.
So, if someone wants to investigate the audibility of differences the main thing is to design an experiment that only investigates the audibility of the difference, not other things the experimenter didn´t think about.
To quote from Bech&Zacharov:
Soren Bech, Nik Zacharov,
Perceptual Audio Evaluation–Theory, Method and Application
Unfortunately, that's irrelevant to the question I raised; the unstated "confounding effects" don't inhibit listeners from detecting the list of often very subtle differences I listed as examples. So the mystery remains.
No, this mystery is related to the question if an experimenter is able to design a valid test or not.
You may remember that i´ve cited JJ several timesm that listeners can reach amazing sensitivity in controlled listening tests if they had training before and if negative and positive controls were used before/during the test.
So the mystery remains mainly why so often invalid or nonobjective and inreliable tests were done in the audio field.
Beside the ITU-Recommendations that address this point, another quote from Bech/Zacharov seems to fit:
So in controlled listening tests 'audible differences' are not detected. On the other hand it has been suggested (by Thorsten) that there could possibly be some statistical evidence that people are detecting differences where there are none. The conclusion must be that controlled listening tests don't work so we should instead use uncontrolled listening tests? Concentration on the task in hand renders that task impossible, so to help listeners hear better we make them aware of price/styling/brand/design/designer and we can be confident that this extra information won't bias the results?Jakob2 said:
Hi,
First, there is both evidence that people under test conditions fail to hear present and audible differences (nocebo) and that hear non-present differences (placebo).
This is hardly news to anyone in touch with developments in subjective testing.
My point about the "much greater than chance when compounded" number of wrong identification in ABX testing is subtly different.
ABX requires correct identification of X as A or B to be considered "right", all else is "wrong". This is opposed to for example presenting A and B in the form of X and Y (where X & Y are always different but it is unknown if they are A or B) and asking listeners to express a preference (which I prefer to use as test).
So I pointed out that due to the implementation of ABX a abnormal number of "wrongs" in fact suggests statistically speaking that a differences existed, but influenced the listener towards wrong identification.
Effectively ABX discards much statistical data, which if correctly evaluated could give secondary indications of differences and it over-emphasises the avoidance of falsely rejecting the null hypothesis for the numbers involved in the tests in the first instance. If we happen to have biased subjects (with all that entails regarding statistics) we have an even greater weighing towards not rejecting the null hypothesis when it is in fact incorrect.
Past that, the specifics of many highly publicised ABX tests reveal a methodology in procuring test subjects ("You must come to me and prove to me what you hear" type challenges) that seems to make a lot of bias is present. Then generally other tests stresses are also made as high as possible.
No control tests are conducted to ensure that the given test set-up and test subject combination is in fact capable to return positive results with known audible differences.
It is this triad of criticisms of which each alone suffices to cast grave doubt on the usability of the results of these experiments, which when taken combined, place the ABX listening chair as means to establish the audibility or not of much of anything in one group with the Mediveal ducking stool as means to establish if a certain red-haired, green-eyed female was in fact a witch to be burned (in which case she would float) or in fact innocent (in which she would not float and hence drown)...
Ciao T
First, there is both evidence that people under test conditions fail to hear present and audible differences (nocebo) and that hear non-present differences (placebo).
This is hardly news to anyone in touch with developments in subjective testing.
My point about the "much greater than chance when compounded" number of wrong identification in ABX testing is subtly different.
ABX requires correct identification of X as A or B to be considered "right", all else is "wrong". This is opposed to for example presenting A and B in the form of X and Y (where X & Y are always different but it is unknown if they are A or B) and asking listeners to express a preference (which I prefer to use as test).
So I pointed out that due to the implementation of ABX a abnormal number of "wrongs" in fact suggests statistically speaking that a differences existed, but influenced the listener towards wrong identification.
Effectively ABX discards much statistical data, which if correctly evaluated could give secondary indications of differences and it over-emphasises the avoidance of falsely rejecting the null hypothesis for the numbers involved in the tests in the first instance. If we happen to have biased subjects (with all that entails regarding statistics) we have an even greater weighing towards not rejecting the null hypothesis when it is in fact incorrect.
Past that, the specifics of many highly publicised ABX tests reveal a methodology in procuring test subjects ("You must come to me and prove to me what you hear" type challenges) that seems to make a lot of bias is present. Then generally other tests stresses are also made as high as possible.
No control tests are conducted to ensure that the given test set-up and test subject combination is in fact capable to return positive results with known audible differences.
It is this triad of criticisms of which each alone suffices to cast grave doubt on the usability of the results of these experiments, which when taken combined, place the ABX listening chair as means to establish the audibility or not of much of anything in one group with the Mediveal ducking stool as means to establish if a certain red-haired, green-eyed female was in fact a witch to be burned (in which case she would float) or in fact innocent (in which she would not float and hence drown)...
Ciao T
- Status
- Not open for further replies.