If I am to invest in expensive cables then I need to be convinced that they sound significantly better, even in a 'non-ideal' test as my listening environment is certainly 'non-ideal' and is subject to change.
Surely the effort here on designing the perfect test is becoming too pedantic - if enough people are involved or repetitions take place then the 'subjective errors' should be 'almost' removed.
One can spend a lifetime arguing about the effectiveness of different tests, but losing sight of the real issue - tends to indicate that any differences that do exist between cables will be extremely hard to prove and only supports the case that there will be no clear trend detected.
Surely the effort here on designing the perfect test is becoming too pedantic - if enough people are involved or repetitions take place then the 'subjective errors' should be 'almost' removed.
One can spend a lifetime arguing about the effectiveness of different tests, but losing sight of the real issue - tends to indicate that any differences that do exist between cables will be extremely hard to prove and only supports the case that there will be no clear trend detected.
keladrin,surely I don't want to spend my life looking for the ideal cable test method.The only way to convince yourself is to listen and if you don't detect any difference then just enjoy your music and don't bother.But the old hi-fi saying that we can mix two wrongs to make a right is no longer acceptable.
soongsc said:
Body language? You know, the same thing that tips off the customs man no matter how hard you try to look inconspicious.
You guys are just groping around. You should read some test events like the one on the BAS website from Stanley Lipshitz.
http://www.bostonaudiosociety.org/bas_speaker/abx_testing2.htm
http://www.bostonaudiosociety.org/bas_speaker/abx_testing.htm
http://www.bostonaudiosociety.org/bas_speaker/wishful_thinking.htm
You don't need to agree, but at least it helps you to skip reinventing the wheel all over again.
Jan Didden
Panicos K said:
reductio ad absurdum
It's just a case of comparing apples to apples. Once a "speaker wire" becomes more complex through the use of added LCR components in the form of a network, which is intended to affect measurable aspects like frequency response, or introduce other types of distortions then it ceases to simply be a "speaker wire", it has more akin to a filter at that point. We're not talking about comparing speaker wire to a device that has been engineered to actually make a difference within audio frequencies. It'd be like comparing 12guage OFC ropelay to a low pass filter.
I would argue that given two metalic conventional cables -- one worth $40 and one "worth" $1400, you are essentially comparing two identical cables....
Aengus said:
Yes, I agree. I'm still reading through them. But I thought the experiments showing even an elevated noise floor was inaudible when music was being played (comparing a ADC/DAC to straight source) was particularly interesting.
It speaks to the early point that although differences may indeed exist, their audibility is often insignificant.
macgyver10 said:
Hi macgyver10. Use a limiting case, my trivial example. Could you differentiate between PBS Alphas and Wilson WAAM monitors in a blind test over the phone? Ridiculous for sure, but it demonstrates that the test protocol can set limits to the ability to discern differences. If you agree to that now it's just a matter of degree. In the case of cables the claim is that a device comprised of 4 solder/crimps, two connector bodies and a paired conductor displays audible differences we don't know how to measure. Saying we've measured the ABX comparator begs the question of whether whatever unknown 'factor x' we're testing for isn't obscured. Even on the face of it and standing back from the contention of this topic I don't see how it can be logically claimed a device like the ABX comparator shown earlier with many connectors, dozens of solders and uncontrolled wire jumpers open to RF and hum pickup, multiple dielectrics and switches is absolutely certain to have no impact when we conceivably don't know what to measure for. (Don't even get me started on soundcard based ABX.)
Won't work if I understand it correctly. Inactive 'Amp A' will terminate the output of 'Amp B' and present a near dead AC short to its output. It's easy to overcome though, use DPDT relays which lift the inactive amp entirely and terminates it into a dummy load resistor. It's not a bad test expanded further. Make the relay switching device similar (or identical) to the muting/protection relays used at the output of any random top-end amp. For argument say an MBL or Krell output relay. Take it one step further and instead of two different amps use two different amp/speaker cable pairs. To argue such a relay device obscures the differences between amp/speaker pairs is to argue it's the predominant auditory determiner of the sound of the Krell or MBL. That's the trick here and with the pre and amp example from earlier. Don't prove amps and cables which measure similarly don't sound different, instead work towards proving any potential differences are at a level obscured by the components used within the device being tested. Logically a different claim, in a practical sense identical and as far as I can tell not open to criticism, other than the amount of work required.
BTW, I totally disagree about not wanting 'live sound' at home. One of the benefits of working in radio is live acoustic performances by up-and-coming artists at the office. This morning Serena Ryder sang and played acoustic guitar ten feet from my door. Trust me when I say that's the sound I want.
Then simply test for the audibility of the switching device, as the Boston Audio Society tests did.rdf said:
Of course, the ultimate subjectivist could carry your predictable argument back in infinite regress, and ultimately make an appeal to Heisenberg's uncertainty principle. Such a person, of course, is not interested in finding out the truth, but only in making excuses.
rdf said:
I don't remember saying that I wouldn't want "live sound" at home? But I very well could have.
If I did, then what I would have meant, is that the example you state above, is NOT what the the vast majority of source material would deliver. It's also not what the vast majority of "live events" deliver. More often it's going to be a ensemble which may or may not be acoustic, partially acoustic, or full out PA'd.
The PA'd concert, is probably not what you want your home system to sound like.
The remaining two are a total crap shoot as to whether or not a properly recorded performance would sound MUCH better. It's been my experience having been to many shows in many different venues in many different cities that it tends to be rare to experience high fidelity "live".
However, that being said, we get into the semantics of what it means to sound "live". Again, rather than accuracy, personal preference takes over.
I want my system to sound "live" as well. But IMHO I tend to feel that dynamics and frequency range, rather than "flatness" of frequency response tends to make my hackles stand up.
But then again, I'm a drummer, and I love percussion. That's an area that demands very dynamic and full range systems, but can suffer from a "lumpy" FR plot. For alot of percussion material, this might not matter, for an acoustic set like you mentioned above, it could potentially sound terrible (or great! who knows?).
Also, as you're probably very aware, the close mic'd drum method most often employed, with myriad compressors and gates, does very little to maintain "live", but it does often make the drumset sound WAY better than it does acoustically. Rare is the acoustic kit, that sounds great. And often a great sounding acoustic kit, is a nightmare to mic and amplify (very resonant).
In the end though, it's going to come down to whether the source material can deliver, no matter how good my system "sounds".
Thanks Jan, for posting those article links. Very good reading!
I think they illustrate nicely how difficult the whole A/B or ABX testing is. P.T. Barnum learned long ago that if you set out to fool people, you will. And don't be easy on the ABX test, it IS set up to deceive you, that's the point. Deceit is not the desired outcome, but it is the result.
If you are forced to find a "difference" between A & B, then tell if X is A or B, you are hunting hard for any clues you might find. The number of false guesses has to go up, simply because you are forcing people to guess. Ask a child a question that he doesn’t know the answer to. You will usually get an answer, often a funny one, just because the child knows that an answer is expected. Art Linkletter made a career of this.
From reading those articles it would seem that obvious differences are easy to spot and the scores reflect it. The pink noise test, the background noise test. But when the difference becomes subtle, the test is no as reliable. Not surprising.
One of the problems of this sort of test might be the "Golden Ears" syndrome. If you have 100 listeners, and only one golden ear, his results will be swamped by those who can't tell the difference, or just don't care.
Let me illustrate this with a non-audio example:
I work in the fine art printing business. We do rather good, expensive reproductions of paintings. An artist brings in a painting, I scan it, we color correct, then print. We do the best we can to get an exact match to the original artwork, a very difficult task. It is not possible to get an exact match, but we sometimes get so close that the painter can not tell our print from the original!
But we can tell. How? We know what to look for. As visually astute as the painter may be, he is not used to seeing prints all day long. We are, and know what they look like.
Often there will be a match print and a proof up on the color correction board because we are checking to see if the current print run matches the master proof. Most people who walk by say the same thing. "Those look exactly alike." But they don't. When we point out the differences, everyone sees them, and they will often go on to find ones we haven't seen. It is all in knowing what to look for. Once the differences are pointed out, they become noticeable. No "Golden Eyes" needed.
They trouble here is that those who haven't been trained to see the differences will not see them again the next time. They forget what to look for.
Sorry to take this away from audio, but I thought it might help make a point.
As useful as AB testing is, it can lead to trouble, as you've all been discussing. Making a change in a system you know well, and with music you know well, can be much more obvious. But it is not without its own problems.
I think they illustrate nicely how difficult the whole A/B or ABX testing is. P.T. Barnum learned long ago that if you set out to fool people, you will. And don't be easy on the ABX test, it IS set up to deceive you, that's the point. Deceit is not the desired outcome, but it is the result.
If you are forced to find a "difference" between A & B, then tell if X is A or B, you are hunting hard for any clues you might find. The number of false guesses has to go up, simply because you are forcing people to guess. Ask a child a question that he doesn’t know the answer to. You will usually get an answer, often a funny one, just because the child knows that an answer is expected. Art Linkletter made a career of this.
From reading those articles it would seem that obvious differences are easy to spot and the scores reflect it. The pink noise test, the background noise test. But when the difference becomes subtle, the test is no as reliable. Not surprising.
One of the problems of this sort of test might be the "Golden Ears" syndrome. If you have 100 listeners, and only one golden ear, his results will be swamped by those who can't tell the difference, or just don't care.
Let me illustrate this with a non-audio example:
I work in the fine art printing business. We do rather good, expensive reproductions of paintings. An artist brings in a painting, I scan it, we color correct, then print. We do the best we can to get an exact match to the original artwork, a very difficult task. It is not possible to get an exact match, but we sometimes get so close that the painter can not tell our print from the original!
But we can tell. How? We know what to look for. As visually astute as the painter may be, he is not used to seeing prints all day long. We are, and know what they look like.
Often there will be a match print and a proof up on the color correction board because we are checking to see if the current print run matches the master proof. Most people who walk by say the same thing. "Those look exactly alike." But they don't. When we point out the differences, everyone sees them, and they will often go on to find ones we haven't seen. It is all in knowing what to look for. Once the differences are pointed out, they become noticeable. No "Golden Eyes" needed.
They trouble here is that those who haven't been trained to see the differences will not see them again the next time. They forget what to look for.
Sorry to take this away from audio, but I thought it might help make a point.
As useful as AB testing is, it can lead to trouble, as you've all been discussing. Making a change in a system you know well, and with music you know well, can be much more obvious. But it is not without its own problems.
- Status
- Not open for further replies.