Golden Ears.

Re: Re: Golden Ears.

It's not unreasonable to ask you to pay for these report - back issues of HiFi+ and other such magazines cost money.

Let's review what the word "published" means.

In the context of Stereophile or HiFi+, the article author writes what the hell he pleases, passes it to the editor, who mindful of advertizing revenue, might suggest some appropriate changes and then, pretty much no matter what is said, it is published without any attempt to verify a single word. Stereophile have a habit of publishing graphs of measurements, but there is usually no attempt to correlate the measurements with the subjective review..

On the other hand: to publish an article in the Journal of the AES or IEEE, requires the article to pass peer-review. The article will be read by at least three independent reviewers who are experts in appropriate fields. The article will only be published if the findings it makes follow from the evidence that is provided. The evidence will also be examined to determine if it has been gathered in an appropriate way (to not introduce bias).

Nothing is "Published" in Stereophile/HiFi+ etc, it's a vanity publication.

And Olive is entitled to draw that conclusion because it fits with the evidence he has presented.

Note however, he does NOT name "cables", "amplifiers" or "CD sources" as nuisance variables.

That's fair enough. But, the inference is usually made the other way around by the buying public - if it looks good, it sounds good. Ever wonder why so much money is wasted on fancy veneers?

We can continue presenting evidence to you of what other people can or cannot hear all day, but it will do no good. Your going to doggedly hang onto to your beliefs about what you can hear.

There is a vast body of evidence about what people (in general) can hear, and the probabilty that you (in particular) are in some way special is vanishingly small

If you can argue the facts - argue the facts;
if you can argue the law - argue the law;
otherwise pound the table.

Keep pounding the table [You're not fooling any of us..]

Now, if you want more evidence (which you have asked me for) you only need to follow up the references in the paper that you already have.

Alternatively, why not phone up your mate Greg Timbers and ask him what the hell his boss (Floyd Toole) was doing wasting his time being president of a "mythical and short sighted organisation" such as the AES

Oh wait, would this be the same Greg Timbers:

"An Analysis of Some Off-Axis Stereo Localization Problems "
Author(s): Eargle, John; Timbers, Gregory
Publication: AES Preprint 2390; Convention 81; October 1986

 

Re: Re: Golden Ears.

It costs $180 to get a year's full access to the online edition of the Journal of the American Chemical Society, and that doesn't even get you physical copies of the journal - that's a tidy $578 per year outside North America

 

Re: Re: Re: Golden Ears.

Dat, try visiting the said publications web sites and see if you can peruse and download the reviews without cost. I beleive you might be surprised.

Agree, with the proviso that Stereophile do try to compare these findings to a small degree, and are guilty of assuming that the reader will have an idea of the speaker's characteristics based upon the subjective and objective reporting.

To buy anything on the strength of a review would be naive to say the least, as would acepting the findings of the Chief engineer at Harman (who I might add manufacture both Infinity speakers and a number of other brands that fail to get the recognition that they may deserve in the audiophile press.

Quite possibly because they were a constant thoughout the tests I don't have the text with me at home, but how were the speakers moved and sited in the test room to insure that the direct and reflected energy was optimised for each at the listening position?

To put the record straight, I don't find the AES to be totally flawed, just some of the methodology used to prepare some of the papers.

 

Michael, Merlin, can you guys pleeeaaase stop trying to have an argument all the time??

Maybe you don't mean to argue, but it does come off that way when reading. Why can't both of you just believe what you want to believe and not worry about converting or 'informing' the other? Its just getting quite tiring to be honest.

Thanks.

 

Re: Re: Re: Re: Golden Ears.

What on earth does that have to do with anything? This was a test of the response of different listening groups, not a speaker review, so as long as the tests were carried out consistently it doesn't make a blind bit of difference whether "the direct and reflected energy was optimised". In fact the test could just as well have been carried out with the same speaker in different positions, although that would have removed the frission of trying to guess which speaker was which

 

Pete,

Wouldn't it be relatively easy to exaggerate the differences in perceived quality by insuring that room related issues affected the performance of some of the units more than others, given the differing dispertion pattern?

I can make Logans sound diabolicle by moving them 3 cm for heaven's sake. They can also sound quite superb. If you create an artificially ideal enviroment for say the Infinity and do not set up all speakers to their optimum, it's totally possible to make it very easy for the test subjects to express a preferrence, regardless of their training or hearing ability.


All I asked for was the details of positioning.

 

Quite possibly, but again irrelevant - all he was doing was asking the subjects to listen to four different audio presentations and rank them. As has been made quite clear, the object of the exercise was to determine whether the different groups perceived the differences in a consistent fashion, which they did (notwithstanding a greater spread of results in the less experienced groups). The nature of the differences between the four things they were asked to compare really isn't relevant

 

But surely the size of the differences are?

 

Not really, as long the differences are within a sensible range such that the listener is called upon to make a judgement (ie. not two speakers wired out of phase and pointing directly at each other, for example) - and given that none of the test systems was given nul points, we can safely conclude that that criterion had been met. As Michael has already said, "what's important is that the trained listeners ranked the speakers in the same order with similar relative scores.".

And in any case, you can't quantify the differences without doing an experiment of this kind in the first place

 

I'm afraid we'll have to agree to disagree there Pete.

It is possible to emphasis differences between the performance of loudspeakers with varying technologies by incorrect setup. Just having the bass unit of one speaker in a null will make a large difference.

Now if you make these differences large enough, surely anyone could hear them, as the test proves. What was missing was a reference, which makes the relative scoring and scale difficult for me to relate to.

I am lost by the time the conclusion comes along. What are they saying? That given enough time, all listeners came to the same conclusion? If so surely that raises more questions than answers? Particularly when it comes to forum members claiming to switch cables and hear no difference, which is where this all started.

 

A couple of points to note.

a) Preference doesn't necessarily relate to accuracy.
The trained listeners may well be able to hear more accurately and come to the same order of preference for the speakers as untrained listeners. The two are not directly correlated.

b) Without knowing the performance differential between the four loudspeakers (The author erroneously correlates price with performance - something everyone here knows isn't true) you can't tell if the trained listeners actually needed to use their increased ability to discriminate than the untrained listeners to come to their conclusion. A trained wine taster can taste subleties that untrained drinkers can not, you wouldn't need a trained wine taster to distinguish between a desert wine and a dry red.

c) The difference between the group numbers is very significant.
It takes 200+ untrained individuals to come to the same conclusion as <20 trained listeners. Well, it would seem to me that
i) This makes no allowance for the distribution of ability within the untrained group
ii) The law of averages is likely to end up with a more accurate result than any given individual one.

Such a disparity would be like saying that no one has above average singing abilty based on the example of a group of 10 trained singers hitting the same notes as a crowd of 200 untrained singers. A large crowd never seems to sing out of tune, it doesn't follow however that everyone in that crowd can sing well, or sing at all.


GTM

 

Some interesting points GTM but nothing to counter what I believe was the key conclusion of the study which was that the training that trained listeners have doesn't bias them (intentionally or unintentionally) to prefer different things from loudspeakers to untrained listeners.

This goes somewhat contrary to the popular belief that Joe Public prefers a sound that emphasises bass and treble (the "boom tizz" sound).

Sure, but two of the loudspeakers (P & I) ranked pretty closely so presumably were pretty close in performance. Both the trained and untrained listeners managed to hear this small difference and both preferred P to the same degree.

Where? He does no such thing as far as I can see . All he says is "Given the relatively high prices of the loudspeakers, they should in theory represent “state-of-art†designs in terms of technical and sonic performance".

Surely anyone would accept that there is a loose price/performance relationship in loudspeakers and that picking 4 speakers from $5000 to $11000 is likely to get you 4 very good loudspeakers. Anyway, the choice of loudspeakers is irrelevant to the test. He could have picked 4 $200 speakers and got similar results. It's interesting to note that the speaker that performed the worst by some margin was the most expensive one.

But that's not relevant to the test. He's not trying to show that every single person can hear equally well, merely that on the whole we all hear the same things and that training just improves our discriminatory abilities.

Yes, he is basically saying that given enough time all listeners come to the same conclusion. It would be no different with cables. If you did the same test with 4 different cables you'd find that no-one, whether trained or not, would be able to hear any significant difference

Michael.

 

What's the old adage?

Give one hundred monkies one hundred typwriters, and in one hundred years they will come up with Hamlet?

Doesn't neccessarily follow that they will be able to differentiate between Keats and Shelley in a quick comparison though does it

 

As I've said it really doesn't matter how the differences between options B, I, P and M arise, whether by bass nulls or intrinsic quality differences between the loudspeakers. Regardless, the relative scores given by each group were basically the same, so insofar as we can quantify these things the groups perceived the same "amount" of difference between the different options. Now, obviously this is hardly definitive proof, but it can only count as evidence against the concept of "golden-eared" types much more sensitive to small differences than Joe Public. I'm not sure I see what you're getting at when you talk about a reference - I can't really see how the scale could have been made any more "absolute", particularly in the absence of a definitively perfect playback solution (although now I write that I vaguely recall someone did try to define an absolute sound quality scale once).

GTM - surely this test is as close as we're going to get to "knowing the performance differential between the loudspeakers". And on the subject of large groups singing out of tune - have you never heard a football crowd on the terraces?

 

The camp is still divided some way in the middle of whatever. Maybe the title of the thread was a bad idea. In retrospect it was too provocative. It should be, This is a boring article. Read it slowly and then let has a debate about it if you like since it is very boring. :rds2: Then again it would makes us all into passionless old git sitting around smoking pipes.

Anyway, it would be nice if we all could read the whole article and then discuss it in a proper way. Could some one summarise what was the author's intention of the study, the method used and then the conclusion. These are presented in pieces but only if it suits their own conclusion. It seems Michealab is the only one who has read the article, tried to present it is a fair manner before forcing his own preconceive idea.

 

There's not really any point in having this debate if you're just going to come up with pointless one-liners of no relevance.

I'm not going to fall into your trap or insult your intelligence by explaining to you why the monkeys story and your extrapolation of it has got no bearing whatsoever on the discussion at hand.

Michael.

 

Oh go on Michael Besides which, I don't really have much to insult so give it a go.

Exactly why does the discovery that untrained listeners take a long time to reach a conclusion that varies widely enough for there to need to be a large pool of said opinions to eventually come to the correct one have nothing to do with my contention that less experienced listeners might struggle to hear subtle differences in quick comparisons at home?

Oh and when are you going to finish that bloody amp

I bet it will sound stunning. Are you going to seek the opinions of others in a properly conducted test out of interest?

 

You've hit the nail on the head right there . Eventually being the key word. Those theoretical monkeys would eventually write Hamlet. It's a question of probability. What they're typing is totally random but there has to be a finite (but mind bogglingly small) probability that any randomly chosen sequence of 1million (or whatever it is) characters will be "Hamlet". If you started the monkeys off today, Hamlet might turn up tomorrow or only in a million years (assuming the monkeys and their offspring can be arsed to continue for that long).

What was happening in that test was not probability or random choice. Sure, if you have a group of listeners choosing randomly they would eventually come up with the same scores as actually occured. The average of their scores would be just as random though and utterly meaningless. They weren't choosing randomly though. The reason a larger number of them was required was simply because there's more variance in their answers (compared to the trained listeners) but on average they gave the same relative scores as the trained listeners.

The test had nothing to do with whether less experienced listeners (or any listeners for that matter) might struggle to hear subtle differences in quick comparisons at home.

Nearly there. I've done all the drilling of holes and now it's just a question of connecting all the bits up. I hope to have it finished this w/e before another 2 weeks in London. I'm going to seek the opinions of others of course. It won't be a properly conducted test though

Michael.

 

How do you explain why people like different speakers, if in a DBT they all choose the same, regardless of experience and personal taste...

 

No it didn't.

But as a byproduct it does suggest that this might be the case