Hi,
So instead we should give up trying to find any correlation between engineering/science and sound quality, and amplifier 'design' and construction should be purely a matter of limited tests not proven to correlate with good sound or pronouncement from a 'scope jockey or Hirsch Houck Labs?
I always smiled at tests, especially math, but in general too...
That happened to me once. But here is the fun part. The test was blind and from what the results looked like EVERYONE ELSE IN THE ROOM heard a difference, which I could not, because it was my modified CD-Player in the test, going up against a stock machine!
Or a cheeky monkey like me claiming to ABX test mains cables when in fact swapping one channels polarity, which the "Audiophiles" in the test found was grossly obvious, while the "Skeptic" who I targeted with this little demonstration showed a random score...
The demonstration was of course that a strong belief to the positive or negative acts as a strong randomising influence in such types of tests with biased subjects and small numbers...
Ciao T
So instead we should give up trying to find any correlation between engineering/science and sound quality, and amplifier 'design' and construction should be purely a matter of limited tests not proven to correlate with good sound or pronouncement from a 'scope jockey or Hirsch Houck Labs?
I always smiled at tests, especially math, but in general too...
That happened to me once. But here is the fun part. The test was blind and from what the results looked like EVERYONE ELSE IN THE ROOM heard a difference, which I could not, because it was my modified CD-Player in the test, going up against a stock machine!
Or a cheeky monkey like me claiming to ABX test mains cables when in fact swapping one channels polarity, which the "Audiophiles" in the test found was grossly obvious, while the "Skeptic" who I targeted with this little demonstration showed a random score...
The demonstration was of course that a strong belief to the positive or negative acts as a strong randomising influence in such types of tests with biased subjects and small numbers...
Ciao T
I agree, given that at present we do not know all the engineering criteria which may be necessary. However, I think we do know some of them so people who design purely by ear (and therefore sometimes make astonishingly poor engineering decisions, such as frequency response which varies with volume control position) and people who evaluate solely by ear are choosing to ignore known facts.Joshua_G said:
No, of course not. As usual you extrapolate from what I said to something I didn't say.ThorstenL said:
So you demonstrate that one particular mains cable sceptic has apparently poor hearing. I hope you are not extrapolating from this to all mains cable sceptics. I am a mains cable sceptic. A few years ago I told a member of my family (a speaker bi-wiring fan) that one channel was inverted - he was quite embarrassed because he (not an engineer) had always felt he had better ears than me yet although he could tell something was wrong he didn't know what it was. Once I told him it became 'obvious'. From this I deduce that at least one mains cable sceptic has better hearing than at least one bi-wiring fan. That is all I can deduce, although I would like to extrapolate to a comment on bi-wiring too.
Those dirty rats! How dare they? Tar and feather them! Run them out of here on a rail! Where are the real scientists? Let them take up arms against this sort of heresy - it should not be tolerated.
I know of no designer of audio gear who designs purely by ear, so I don't see any practical relevance of your comment.
Moreover, any amp with gross poor engineering decisions, like change in frequency response with different volume control positions cannot have an excellent sound quality. So, again, I don't see the practical relevance of your remark.
A solid performing audio gear cannot be attained without the design being based on solid engineering practices to begin with. Listening evaluations come to enable all the fine modifications, beyond and above the basis of solid engineering practices.
As for evaluating sound gear for my own use.
As I wrote above, neither published technical data nor magazines reviews can foretell me the sound quality of any piece of audio gear. So all I'm left with is my own hearing. Now, all my audio gear is commercial, with published technical specifications, so I know there are no gross technical abnormalities, like uneven frequency response, or too high distortion level, or noise. But also audio gear having much inferior sound quality have no such abnormalities.
I plan to build a pre-amp after the late Allen Right's RTP3C. For sure I'll test it for frequency response, noise and distortion. But the measurements wouldn't substitute listening tests, which will follow the measurements.
Furthermore, when I'm searching for the best position of my loudspeakers, I use both music and a CD containing sweep tones over the audio range. This is in order to locate certain frequencies dips and peaks, which are difficult to note with music alone.
Hi,
I did not extrapolate, I simply inverted your statement to illustrate that if we take things as you write, black/white, we logically end up there and hence neither extreme is acceptable.
We have the Heglian "Thesis/Antithesis" conflict at work, which can only be overcome by synthesis.
I would suggest that if one side stops to insist that the other first must pass their questionable test (see below) to be taken serious, the other may be more willing to consider working in serious testing, so that in the synthesis of "this measures this way" and "this sounds that way" real progress may be made.
Up to now I find that "synthesis" happens only with a few rare individuals, all of which tend to post significant achievements in "the art of sound" as a result. To my mind come readily names like Nelson Pass, John Curl, Charles Hansen and Ed Simmons, to limit ourselves to members of this club.
I also find that many others who have posted significant achievements in "the art of sound" come from a very much subjectivist position and approach, while I cannot really say the same thing about the opposite position to any material degree.
It would seem that "measures good enough" rarely results in accolades for the sonic performance, though many attempts to just make measured performance worse (add distortion etc.) have equally failed to gain said accolades.
No, his hearing is quite good and normally he CAN hear polarity inversion in one channel (almost everyone can, honest).
What I demonstrated was that if such a generally good Listener (a former BBC Engineer who was quite put out by my little caper I may add) has been sufficiently "primed" or "biased" he will react to a readily audible stimulus in a manner that is indistinguishable from chance under the ABX protocol (though more appropriate statistics may reveal something), yet unbiased listeners can identify the stimulus correctly 100% of the trials.
I agree that we should not necessarily extrapolate from this to presume no cable skeptic would hear such a difference in a similarly arranged test, nor that every cable believer would always hear the present difference (I'd give it pretty good odds though).
Where this does have a bearing though is the precise ABX test Messers Clark/Krueger/Nousaine/Lipshitz seem to excel at, the "Challenge", except inverted, that is directed to reveal the hidden bias in ABX.
In this case the challenge was presented to the subject to come hear the massive and certain difference mains cables made, which he accepted, "knowing" that such differences cannot exist (something I will agree with, in many cases such differences are subtle by far more than the stimulus I actually employed).
When he was presented with a stimulus that should have been readily audible but believed that this stimulus was one that "cannot be" his belief became his "reality", a reality completely divergent from everyone else in the room.
So, we MAY extrapolate that being human the "golden-eared audiophiles" - being targeted by the ABX Mafia for first abuse to make them react to the bait, then the musical chair ("you see we have this here ducking stool") and then more ridicule when predictable results ensue - are likely liable to the same bias, in other words they are liable to hear differences where non exist...
A POSSIBLE hint is that in surprisingly (from a statistical viewpoint) many ABX tests the tests subjects tend to score WORSE than pure chance.
Now ABX concludes the listener was not able to identify the difference, which from a statistical science viewpoint amounts to a deliberate manipulation and very poor statistics, but that never stopped an ABX Fanatic.
The ABX Mafia then declares this insignificant as for each individual tests it falls within expected variances for chance (which it does - tough if you aggregate several such tests the significance and statistical power changes considerably), ignoring the hint that if you have many such results they themselves compound to suggest that a stimulus exists which biases subjects to be more wrong than pure chance (whatever this may be).
BUT, ABX was never about finding out the truth of the matter, so the failure to take such hints should not surprise.
In the end there are, as my Statistics Professor used to say lies, damn lies and then there are statistics.
I would add the corollary that one of the most oedius and perfidious abuses of statistics is called ABX, in my not so humble opinion...
However, no matter what arguments one may supply to illustrate the major problems with ABX, there will always be those who will not be swayed by evidence, proof but who will continue to exercise blind faith, instead of using their own grey matter...
I believe that I have abused John's hospitality in this thread about his design sufficiently to risk overstaying my welcome. So lets return to discussing the blowtorch, which I do find interesting, as all this debate on ABX testing invariably ends up circular and is already incredibly boring.
Ciao T
I did not extrapolate, I simply inverted your statement to illustrate that if we take things as you write, black/white, we logically end up there and hence neither extreme is acceptable.
We have the Heglian "Thesis/Antithesis" conflict at work, which can only be overcome by synthesis.
I would suggest that if one side stops to insist that the other first must pass their questionable test (see below) to be taken serious, the other may be more willing to consider working in serious testing, so that in the synthesis of "this measures this way" and "this sounds that way" real progress may be made.
Up to now I find that "synthesis" happens only with a few rare individuals, all of which tend to post significant achievements in "the art of sound" as a result. To my mind come readily names like Nelson Pass, John Curl, Charles Hansen and Ed Simmons, to limit ourselves to members of this club.
I also find that many others who have posted significant achievements in "the art of sound" come from a very much subjectivist position and approach, while I cannot really say the same thing about the opposite position to any material degree.
It would seem that "measures good enough" rarely results in accolades for the sonic performance, though many attempts to just make measured performance worse (add distortion etc.) have equally failed to gain said accolades.
No, his hearing is quite good and normally he CAN hear polarity inversion in one channel (almost everyone can, honest).
What I demonstrated was that if such a generally good Listener (a former BBC Engineer who was quite put out by my little caper I may add) has been sufficiently "primed" or "biased" he will react to a readily audible stimulus in a manner that is indistinguishable from chance under the ABX protocol (though more appropriate statistics may reveal something), yet unbiased listeners can identify the stimulus correctly 100% of the trials.
I agree that we should not necessarily extrapolate from this to presume no cable skeptic would hear such a difference in a similarly arranged test, nor that every cable believer would always hear the present difference (I'd give it pretty good odds though).
Where this does have a bearing though is the precise ABX test Messers Clark/Krueger/Nousaine/Lipshitz seem to excel at, the "Challenge", except inverted, that is directed to reveal the hidden bias in ABX.
In this case the challenge was presented to the subject to come hear the massive and certain difference mains cables made, which he accepted, "knowing" that such differences cannot exist (something I will agree with, in many cases such differences are subtle by far more than the stimulus I actually employed).
When he was presented with a stimulus that should have been readily audible but believed that this stimulus was one that "cannot be" his belief became his "reality", a reality completely divergent from everyone else in the room.
So, we MAY extrapolate that being human the "golden-eared audiophiles" - being targeted by the ABX Mafia for first abuse to make them react to the bait, then the musical chair ("you see we have this here ducking stool") and then more ridicule when predictable results ensue - are likely liable to the same bias, in other words they are liable to hear differences where non exist...
A POSSIBLE hint is that in surprisingly (from a statistical viewpoint) many ABX tests the tests subjects tend to score WORSE than pure chance.
Now ABX concludes the listener was not able to identify the difference, which from a statistical science viewpoint amounts to a deliberate manipulation and very poor statistics, but that never stopped an ABX Fanatic.
The ABX Mafia then declares this insignificant as for each individual tests it falls within expected variances for chance (which it does - tough if you aggregate several such tests the significance and statistical power changes considerably), ignoring the hint that if you have many such results they themselves compound to suggest that a stimulus exists which biases subjects to be more wrong than pure chance (whatever this may be).
BUT, ABX was never about finding out the truth of the matter, so the failure to take such hints should not surprise.
In the end there are, as my Statistics Professor used to say lies, damn lies and then there are statistics.
I would add the corollary that one of the most oedius and perfidious abuses of statistics is called ABX, in my not so humble opinion...
However, no matter what arguments one may supply to illustrate the major problems with ABX, there will always be those who will not be swayed by evidence, proof but who will continue to exercise blind faith, instead of using their own grey matter...
I believe that I have abused John's hospitality in this thread about his design sufficiently to risk overstaying my welcome. So lets return to discussing the blowtorch, which I do find interesting, as all this debate on ABX testing invariably ends up circular and is already incredibly boring.
Ciao T
Joshua,
You may wish to add MONO pink noise to your arsenal of test signals.
I find that using both in phase and out of phase versions of this are incredibly telling of many parameters, especially where spatial hearing is concerned. If you have some experience of pink noise without response distortion (some headphones can help) it also becomes easy to hear even very narrowband problems...
Ciao T
You may wish to add MONO pink noise to your arsenal of test signals.
I find that using both in phase and out of phase versions of this are incredibly telling of many parameters, especially where spatial hearing is concerned. If you have some experience of pink noise without response distortion (some headphones can help) it also becomes easy to hear even very narrowband problems...
Ciao T
That may or may not be true, although some evidence to support this assertion would be nice. One could ask whether avoiding ABX is also about avoiding the truth of the situation?ThorstenL said:
If golden ears are disposed to find differences when none exist in ABX trials, which you suggest as a possible inference, then it is conceivable that such people may also find differences where none exist in other situations too. If an individual's income depends in part on hearing differences then this will encourage the hearing of differences.
I take your point about the possible value of meta-analysis - combining data from different trials. It would improve the statistics. What if it confirmed that people really are doing worse than chance?
Tests "do you hear the difference" remind me such pictures:
It can be fun, but when I evaluate something, I pay attention on some particular flaws that I want to minimize. Or, when I want to tell how well it sounds in common, I close eyes and try to imagine if sounds are real. The harder is it to imagine, the worse is reproduction.
An externally hosted image should be here but it was not working when we last tested it.
It can be fun, but when I evaluate something, I pay attention on some particular flaws that I want to minimize. Or, when I want to tell how well it sounds in common, I close eyes and try to imagine if sounds are real. The harder is it to imagine, the worse is reproduction.
positive controls, training are part of good perceptual testing experimental design - there have been many amateur tests with bad design - there's no reason you can't AB/X and use multi-dimensional "focus" questions: "do you hear a difference in "X" quality" that the subject may have just been trained for with exaggerated examples, scores only accepted if discrimination on controls exceeds some threshold
when I argue from Clark's AB/X jnd frequency response threshold curves it not to "prove" someone "can't hear that"
my criticisms are based on people crowing about hearing differences when they haven't controlled frequency response level to below those thresholds – no one should surprised at all that they hear differences – I just don't feel compelled to accept their interpretation of "cause"
I am curious why the Carver Challenge can be easily dismissed - Stereophile's reviewers knew which amps were being compared, were listening to their choice of speakers, in their testing room, their choice of source
as professional reviewers they should be familiar with making comparisons - so the "test stress" claim seems weaker - it is their day job to make decisions "by ear" on "sound quality"
when I argue from Clark's AB/X jnd frequency response threshold curves it not to "prove" someone "can't hear that"
my criticisms are based on people crowing about hearing differences when they haven't controlled frequency response level to below those thresholds – no one should surprised at all that they hear differences – I just don't feel compelled to accept their interpretation of "cause"
I am curious why the Carver Challenge can be easily dismissed - Stereophile's reviewers knew which amps were being compared, were listening to their choice of speakers, in their testing room, their choice of source
as professional reviewers they should be familiar with making comparisons - so the "test stress" claim seems weaker - it is their day job to make decisions "by ear" on "sound quality"
Last edited:
Stop it Wavebourn! You are trying to trick us with your filthy A/B tests. I will not fall for it! Reality is much deeper than this.
Stop it Wavebourn! You are trying to trick us with your filthy A/B tests. I will not fall for it! Reality is much deeper than this.
Right, finding differences between two realities is even more fun than comparing two pictures.
Well said and done, T. NO, you have NOT abused this thread in any way. This is the point of keeping this thread going. In the beginning, someone wanted schematics and info about the Blowtorch preamp. I decided NOT to publish the schematics, but instead, help people understand the underlying concepts of the Blowtorch preamp design, as well as hints as how we did it, and how others could do their version as well. This went in that direction for a time, and finally, (years ago) we virtually ran out of specific items on the Blowtorch, so we have extended it to virtually all solid state analog design.
Between attacks, smears, and what I call 'P'ing on the thread' we have made a lot of progress. Let us keep up the good work.
Between attacks, smears, and what I call 'P'ing on the thread' we have made a lot of progress. Let us keep up the good work.
Hi,
I decline further discussion as it would be off topic.
Please start a new thread to discuss ABX and it's manifold shortcomings, if you would like to do so.
Ciao T
I decline further discussion as it would be off topic.
Please start a new thread to discuss ABX and it's manifold shortcomings, if you would like to do so.
Ciao T
Hi Thorsten,
In my mind you also belong to that group.
This is because:
a. I own and enjoy your "second best" commercial CD Player, the AMR CD-777.
b. I follow your approach and philosophy for quite a while and I learned to highly appreciate it.
Thank you, I will. All those options are included in the Test CD supplied with the AMR CDP.
In my mind you also belong to that group.
This is because:
a. I own and enjoy your "second best" commercial CD Player, the AMR CD-777.
b. I follow your approach and philosophy for quite a while and I learned to highly appreciate it.
Thank you, I will. All those options are included in the Test CD supplied with the AMR CDP.
i own a sharp dvd player, and when i play "Norah Jone"s thru my player thence thru my 6LU8 8 watter pp amp, and finally thru my Studio Acoustics tower speakers, i could swear as if she's in my living room playing the piano......
...and I own a Sony DVP-S725D DVD Player - £16.50 from ebay, perfect condition and also a Sony DVP-S7700 DVD Player both of which I use as a transport for CDs into a reasonably good DAC....again excellent. FAR better at the spinning job than most good commercial CD only transports! They allow the music to breath.
Maybe what we need, so that all may agree, is a true _Innocent Eye Test_:
http://www.101bananas.com/art/innocent3.jpg
Thanks,
Chris
http://www.101bananas.com/art/innocent3.jpg
Thanks,
Chris
- Status
- Not open for further replies.