It is an excellent response, if you want to deliver graduates that are worth anything.
Give a man a fish and he has a good meal. Learn him how to catch fish and he has meals for the rest of his life.
Jan
...will spend all day in his boat, drinking.
(Pedant: "teach," not "learn," like "lehren" vs "lernen")
Yes, Jan, but this presupposes two things, actually 3.
1 - that the professor actually is in possession of the information
2- that the "student" does not
3- that the "answers" actually exist and have been discovered
We are not in a school, the assumption of what I have read or not read may or may not be valid, I am not a "student" and this is not a "school" and no one here (for the purposes of this forum, no matter what their day job is) is a "professor".
Item 3 is what the discussion revolves around.
And, as we both know "science" in the past has been either incorrect or incomplete, supplanted by the building of information, knowledge or outright 'out-of-the-blue' revelatory breakthroughs.
And, as I said a few times already, if this is all understood and reduced to practice then come with the specs or lacking that examples of commercial gear that meet the criteria for being "blameless" and indistinguishable from each other.
That ought to be fairly simple, especially given the extent of the studies and science thus far performed!
Or, wait, maybe the DBT and other tests tell us not what is as much as what is not?? But here I speculate. I don't know what the conclusions are. What are they? Surely there must be a general consensus, a trend if not hard solid standards?
I am not saying that there is no value in the research. I am asking what does the research/testing say that is practical and applicable WRT the questions I am asking?
How to apply that research in a practical way. Generally speaking.
1 - that the professor actually is in possession of the information
2- that the "student" does not
3- that the "answers" actually exist and have been discovered
We are not in a school, the assumption of what I have read or not read may or may not be valid, I am not a "student" and this is not a "school" and no one here (for the purposes of this forum, no matter what their day job is) is a "professor".
Item 3 is what the discussion revolves around.
And, as we both know "science" in the past has been either incorrect or incomplete, supplanted by the building of information, knowledge or outright 'out-of-the-blue' revelatory breakthroughs.
And, as I said a few times already, if this is all understood and reduced to practice then come with the specs or lacking that examples of commercial gear that meet the criteria for being "blameless" and indistinguishable from each other.
That ought to be fairly simple, especially given the extent of the studies and science thus far performed!
Or, wait, maybe the DBT and other tests tell us not what is as much as what is not?? But here I speculate. I don't know what the conclusions are. What are they? Surely there must be a general consensus, a trend if not hard solid standards?
I am not saying that there is no value in the research. I am asking what does the research/testing say that is practical and applicable WRT the questions I am asking?
How to apply that research in a practical way. Generally speaking.
Depends on what, and how much of course.
Good point about the 'don't know' response...is this choice available in DBTs ?.
Yes of course, but getting the same/similar descriptions every time kinda renders formal DBT moot in this case....for now.
My subjects typically are honest normal everyday people with no ear training, preconceptions etc...and so far except for two trials, 100% detection and matching descriptions, and preference too.
The two had preconceptions, so are not valid subjects, therefore I discount them from the data set of many hundreds of trials.
Dan.
Q. Whatnoiseannoysanoyster ?.
Could be a cup of dope* (or really large annular hickeys. Brasil Sucks, Man)
*tea leaves for British athletes
Which, of course leads to the natural and logical extension of the idea that all of this hi-fi perception stuff is a done deal.
IF IT WAS a "done deal" then surely there would be a "standards committee" set up that would have proposed a set of "standards"? You know, that say that it's sound is undetectable, or whatever you want to call it... passes DBTs & ABXs with a statistical probability of <fill in the blank>.
In which case, all the mfrs would be trumpeting "our new ultra blowhard unit meets and exceeds the blankety-blank standard"!!
Gee, wonder why this does not exist?
Go ahead and tell me there is no need... is that going to fly as a position now?
In fact I demand such a standard so that the consumer will no longer be misled!
Those who know about this matter can here and now begin to propose the nuts and bolts for the proposed standard(s). We can bring it to the AES or other body after we get the basics laid out. Given the expertise on board here, after a bit, we can spin it off into a new thread... right?
Who is not in favor??
IF IT WAS a "done deal" then surely there would be a "standards committee" set up that would have proposed a set of "standards"? You know, that say that it's sound is undetectable, or whatever you want to call it... passes DBTs & ABXs with a statistical probability of <fill in the blank>.
In which case, all the mfrs would be trumpeting "our new ultra blowhard unit meets and exceeds the blankety-blank standard"!!
Gee, wonder why this does not exist?
Go ahead and tell me there is no need... is that going to fly as a position now?
In fact I demand such a standard so that the consumer will no longer be misled!
Those who know about this matter can here and now begin to propose the nuts and bolts for the proposed standard(s). We can bring it to the AES or other body after we get the basics laid out. Given the expertise on board here, after a bit, we can spin it off into a new thread... right?
Who is not in favor??
Ok, my knowledge of designing DBTs is null, but was thinking about addressing the concern that by accepting a DBT null test result we are essentially throwing away ~50% of the population under test, some of which MAY actually hear a difference.
Would it make sense after a null result to separate the ~50% positives and set up another identical test for them? If the new DBT test result is also null, we can conclude (with a certain probability) that the original ~50% positives were the result of a lucky guess, so the original conclusion (null) gets more weight? And if the new DBT test result are strongly skewed toward a positive, then we may conclude that the original result was due to a special ability. This schema can be repeated until the population sample fails below the statistical significance threshold calculated for the particular experiment.
Of course, numbers have to be attached to such a schema to define what "strongly skewed", "null" for the second test, etc... exactly means, I'm talking about the principle.
Would it make sense after a null result to separate the ~50% positives and set up another identical test for them? If the new DBT test result is also null, we can conclude (with a certain probability) that the original ~50% positives were the result of a lucky guess, so the original conclusion (null) gets more weight? And if the new DBT test result are strongly skewed toward a positive, then we may conclude that the original result was due to a special ability. This schema can be repeated until the population sample fails below the statistical significance threshold calculated for the particular experiment.
Of course, numbers have to be attached to such a schema to define what "strongly skewed", "null" for the second test, etc... exactly means, I'm talking about the principle.
Waly, suppose that some percentage of participants is truly Golden Eared. Then the remainder will show random results. So in aggregate, the group will than show a positive result, which can be analyzed for significance.
Yes, increasing the number of trials can get you more certainty.
And you are assuming a certain format, suitable for answering some questions but not others. It's a distressingly common error to try to have a one size fits all approach- the first rule of good experiments is to define the question being asked, then design the experiment to answer that specific question. You may not be intending to do that, but some people very strongly advocate a Procrustean approach.
Yes, increasing the number of trials can get you more certainty.
And you are assuming a certain format, suitable for answering some questions but not others. It's a distressingly common error to try to have a one size fits all approach- the first rule of good experiments is to define the question being asked, then design the experiment to answer that specific question. You may not be intending to do that, but some people very strongly advocate a Procrustean approach.
Sorry, I just meant honesty (no cheating) so the listener really does not know which is which.
IMO something like a vinyl vs. CD comparison can not in the absolute sense be "blind". A possible one: Mr. X picks a favorite LP that he very familiar the only condition is that it is in print. A disinterested party buys 10 copies of fresh ones and without opening them de-magnetizes 5, randomizes them, and sends them to Mr. Y to open them and listen with Mr. X in any way he chooses.
Maybe one can remove the secret foil from a Bybee thingy keeping it indistinguishable from a complete one. Since it works by simply laying on of the magic there would be no mechanical manipulations of the equipment or ABX boxes involved to cloud the issue.
If you made some real and some fake dongles that were visually indistinguishable and gave a random mix out you could even do sighted tests.
Actually, that IS a blind test, and a very useful one for that sort of question.
I don't think the focus of this discussion should be so much on DBT, as about how to train and otherwise prepare subjects to perform their best during DBT. That's because I think learning is important to hearing some things. Otherwise, you would need a large enough population of random subjects so as to guarantee a statistically significant number of self-trained subjects who learned to hear without looking.
Go right ahead.
What did he find?
I presume you know about D.E.L. Shorter - a fellow countryman, years before Geddes??
And, btw, Billy, do you think that I need you to tell me what this site is about?
Let's see, you arrived here when? I arrived here when? I may have some clue.
I want to remind people that the entire CD making/jitter issue was essentially denied as being in existence intitially - it's underlying source was discovered by some very smart people via a scientific inquiry - because some crackpots thought they were hearing something wrong.
Dismissing everything as "crackpots" and "impossible" is not the best approach.
And, fyi, Billy, many of the people you are posting along side of actually WORK in the very commercial business side of audio although you seem to think that this forum exclusively about people playing with abstractions, doing no listening and hearing nothing being different than the next... so what better place to start such a discussion about "standards".
Are you not in favor, in the abstract sense at least, of such a standard being set??
Has been done. And published
. Results were, shall we say, interesting. And completely ignored, of course.From the abstract:
"Few people will disagree that the strict technical performance of the vinyl LP audio reproduction system has since long been eclipsed by other (digital) storage and replay systems. Yet, vinyl enjoys a sort of renaissance, and many listeners comment favorably on its sound quality. Realizing that ‘hearing’ is not the same as ‘perceiving’, Mike Uwins decided to find out what it is exactly that attracts listeners to vinyl – is it indeed the sound quality or are other factors involved.
Mike devised a very clever system that enabled him to test listener preference in a double-blind setup. His conclusion: there are indeed other factors involved in addition to the sound quality itself, and if these factors are removed, listener appreciation for vinyl is altered."
Jan
- Status
- Not open for further replies.