Electrical measurements of frequency response, distortion, phase are not "objective, reliable and valid"? Or perhaps you meant to refer to audio tests that don't involve instrumentation.
Just curious. What specifically would be a "positive control" for a test of the audibility of two cables?
That's question I finally gave up asking, after being ducked and dodged repeatedly. "Positive control" is apparently a mantra, not a reality. There IS no appropriate positive control for cable effects other than frequency response and level, since there has been no demonstration of listener sensitivity to other aural effects.
Sy, i gave quite specific examples for positive controls, presented arguments about controls and some thoughts where the problem in choosing controls is, so if you have problems to follow arguments you should simply ask specific questions to solve that.
@ boconnor,
sorry, i thought it would be self-explanatory (given the context) that i was talking about listening tests.
But, more generally, the requirements are the same for every test (a test has to be objective, reliable and valid) and using accurate measurement equipment does not ensure that the requirements are met.
In a listening test the experimenter has to rely on the detection capabilities of the participants.
Using controls on sufficient sensitivity levels gives some information about these detection capabilities.
If you don´t have a clue which parameter of a DUT is the specific reason for a (maybe existent) audible difference, but know that the measured difference is small, then it is a good idea to look for the keenest listener.
Small differences already detected by others in DBTs present a good starting point for controls; Paul Frindle gave some examples in his lecture from 1997:
Frindle, Paul. 1997. Are We Measuring the Right Things? Artifact Audibility Versus. Measurement.
AES Conference:UK 12th Conference: The Measure of Audio (MOA) (April 1997)
Paper Number:MOA-05
(abraxalito maybe has some memories on this, because it related to Sony Oxforf)
Another example would be the difference between cd-players like the DUTs used at the ABX homepage.
If there is enough time and patience, it is very interesting to use positive controls on different sensitivity levels to gain some insight into possible learning curves of participants.
Wishes
Last edited:
No, no-one that's been tested by someone else. This is just speculation on my part.
Yes, though given the rather polarised nature of the debate here, I wouldn't expect the subjectivists to be over-ready to make public their experiences. I would be interested in learning (by PM) from any subjectivist who has taken, and failed a DBT about his/her experience. Confidentiality of name will be assured.
Thanks for sharing your experience. I'm sure there's a wide variety of different experiences, as you suggest. What's also obvious is that the 'objectivists' do taunt the 'subjectivists' by saying they're cowards for not submitting to the test. That's been evident from postings on this thread, so its an emotionally charged issue.
Don't see why that would be, so long as the feedback is given after the choice. If it was given before, it would become feedforward
No, you didn't. You have suggested no appropriate "positive controls" whatever. Please remember, a positive control is a stimulus of the thing to be detected at thresholds above known detection levels. All you could come up with was frequency response, which is NOT the thing to be detected.
Oooh, this is already beginning to sound a bit arbitrary. Audible reality is resting on this experiment being controlled, so we do need to have it thoroughly specified.
My sinking feeling is deepening further... This can't reasonably to be held to be something which is the foundation of what's audible and what's not.
I meant how many instances of X being allocated to A or B. Or alternatively, how many button presses by the listeners. Oh and how many listeners?
Seems a bit arbitrary too. Why 95%?
Incidentally, the difficulties in controlling the test do not end with the details I've asked about so far, there are plenty more...
No, I don't mind in the slightest, but you've distorted the meaning in so doing.
Of course the word 'or' doesn't mean the same as 'and' or 'with' when I use it, Humpty not withstanding. But clearly there's a difference between 'instrument or measurement' and 'measurement'.
T9, 10'4, and now catch 22. what's with all these numbers suddenly
Don't we do that everyday? (have evidence of it's collapse) this keyboard feels pretty real, and I am reasonably sure it was real before I ever touched it.
But touch is a perception - according to Steve, what we perceive is not necessarily real, so we'd need to do some experiments to check that your keyboard is indeed as solid as your perceptions are claiming it to be. So I'd say once we allow perceptions to be unreliable and over-ridden by an arbitrary experiment, how do you know? Even with an experiment, the experimenters are dependent on their (potentially unreliable) perception, so it turns out there's no reliable grounds for our science at all.
As I understand it, QM has been one of the most phenomenally successful theories ever, in terms of useful things that have come out of it. Do you have evidence otherwise?
Its useful in the sense that life is known to depend on it. Can't imagine being more useful than that.
I quite agree. So how does, say, Keynesian economics stack up? I mean, even that is useful to bankers.
I'm fairly sure I read the paper when it came out, even though I'd long ago left the then Oxford Digital. I have also had highly stimulating chats with Paul Frindle on subjectivist vs objectivist issues. Do you have a copy of the paper you could send me?
I'm glad you raised the subject of constructivism - that is indeed my operating philosophy of perception. What leads you to cite 'practical issues' for mixing the subject of audibility with constructivist epistemology? It seems to me that they're already intertwined so closely that the difficulty would come in trying to un-mix them.
jakob is right, we have gone over this a few times, time flies so have no idea when it was now!
unfortunately, jakob has not managed to convince me that we need positive controls. (even tho he has been very helpful recently answering questions I asked John...
)I can well imagine we'd need positive controls if we were trying to ascertain JNDs (ha, just pretending I know what I'm talking about by using high falutin important sounding words...really I don't know much about this side of things).
But with Tom (say), it is very clear what we are testing. Tom's claim that he can ascertain the difference between cable A and cable B using only his ears. Tom will choose the two cables himself, and use his own system.
Does that not carry it's own set of controls in a way? We don't need positive or negative controls, test the Fr of his hearing or any such. Either he hears it as he claims, or he does not.
It is unimportant to 'find out how good his hearing is on hearing tests' is it not? Cause that is not what we are testing.
Ok. Think I'll bow out of the thread for a while. Dunno what's going on, think I'm starting to **** people off which was never my intent.
Yes, clearly there is a difference between the two statements, one is about two inches long, the other only an inch.
Should I have said 'without a measurement having occurred'?? Should I have said 'without a record being made'?? (Ie the crater in the ground.)
Of course perceptions are unreliable. I can well imagine the perception of this keyboard might be very different on acid . Solid?? Is it not 99% empty space?
Nope. None. I think it actually is claimed to be the most successful theory full stop.
The point being that it IS useful. And is successful.
Which is a lot more than could be said for 'how can you even prove another person is sentient'. Whilst a lot of fun to muck about with, why I soon grow tired of it is that it goes no where.
It devolves into you cannot prove anything, even that statement itself (or disprove, depending on your bent at the time).
So fun for a six year old, for a time, but of no real use or import in life.
I would not know actually. If pressed, I'd say about the only thing worse than a bamboo sliver under my fingernails might be economic theory.
Hang on, forgot about jazz.
jakob is right, we have gone over this a few times, time flies so have no idea when it was now!
unfortunately, jakob has not managed to convince me that we need positive controls. (even tho he has been very helpful recently answering questions I asked John...
I can well imagine we'd need positive controls if we were trying to ascertain JNDs (ha, just pretending I know what I'm talking about by using high falutin important sounding words...really I don't know much about this side of things).
But with Tom (say), it is very clear what we are testing. Tom's claim that he can ascertain the difference between cable A and cable B using only his ears. Tom will choose the two cables himself, and use his own system.
Does that not carry it's own set of controls in a way? We don't need positive or negative controls, test the Fr of his hearing or any such. Either he hears it as he claims, or he does not.
It is unimportant to 'find out how good his hearing is on hearing tests' is it not? Cause that is not what we are testing.
This will be an ABX test right?
Sure, and SY's protocol is valid and follows known good practice.
The stats experts will have to help you there. My memory of stuff I read was that 16 trials was considered a reasonable number to get the stats working correctly but thats just my fallible memory.
As to number of listeners, panels are good to get the numbers up for the stat calculations. But a well designed and implemented DBT is still valid with a single listener.
Well, its the generally accepted threshold from other fields of research. Its been shown to be pretty robust - in the sense that if you claimed a result with p < 0.05 you are unlikely to be shown up later by other researchers demonstrating that your conclusion was invalid. Of course the lower the figure the better. If you can get p < 0.01 and so forth you're building more and more support for your conclusion, as its less and less likely the result is due to chance.
Its been many years since anyone managed to **** me off, but I'm not finding this as interesting as it was at the start, for sure.
Depends on font size, monitor etc. But you knew that already did you not?
I find discussions where I put words into my interlocutor's mouth to be rather tedious, so I'm not going to say anything about what you 'should' have said. Just be aware that measurement isn't the same as record, as you'd claimed.
I'm not disputing that perceptions can be unreliable, I'm only disputing checking their reliability with an ill-defined experiment and overruling them on a whim as Steve is suggesting.
This is just an example (as I pointed out below) that utility is not an absolute, its depends on who its being useful to. To me this observation you've cited is jolly useful - just as Dirty Harry found the phrase 'I know my limits' to be useful.
Proof exists only in mathematics. But I have the feeling I've said that before....
No real use to you, sure. You do seem to have this habit of over generalising from the specific cases you encounter.
And 'valid' means what here? Surely it doesn't mean 'eliminates all unconscious bias' ? 'Known good practice' is known to whom?
The stats experts will have to help you there. My memory of stuff I read was that 16 trials was considered a reasonable number to get the stats working correctly but thats just my fallible memory.
As to number of listeners, panels are good to get the numbers up for the stat calculations. But a well designed and implemented DBT is still valid with a single listener.
Oh that word 'valid' once again. I do hope you elucidate on it, to me its starting to bear some of the hallmarks of a weasel word.
Well we're dealing with what constitutes audible reality here, so 'unlikely' just isn't rigorous enough.
Incidentally, did you read my earlier comments about potential noise modulation effects owing to RF ingress? Do you consider that they might influence cable audibility and if not, why not?
Gentlemen, are you sure that you are ready to accept the truth about modern electronics?
Dr. John Diamond wrote article about hazard of digital sound in 1980 (!, before compact disk's arrival):
"Human Stress Provoked by Digital Recordings".
So what?! Nothing happened. "Audio industry" ignored the problem that time and still ignores it today (the $$$ sum of possible lawsuits is too high to admit the evil of digital sound).
The problem with cables is the same as with transistors vs. tubes : skin-effect.
I can explain, but I dont want to. Actually I have been explaining it for more than 15 years, and now I am too tired.
You (the society) simply do not like the truth.
Dr. John Diamond wrote article about hazard of digital sound in 1980 (!, before compact disk's arrival):
"Human Stress Provoked by Digital Recordings".
So what?! Nothing happened. "Audio industry" ignored the problem that time and still ignores it today (the $$$ sum of possible lawsuits is too high to admit the evil of digital sound).
The problem with cables is the same as with transistors vs. tubes : skin-effect.
I can explain, but I dont want to. Actually I have been explaining it for more than 15 years, and now I am too tired.
You (the society) simply do not like the truth.
Here's the answer...
So why should anyone believe there IS an audible difference in cables? Because THEY said so, according to JC, abrax..., curly, Monster Cable, MIT, etc. etc. what with all the hand waving and anti-multivariate design whining, not to mention add copy proclaiming all manner of irrelevant and un-demonstrated physical attributes and processes contributing to said difference...
OK.... I'm not convinced. My system is so good that changing cables has no effect on what I hear, even if I know that I (or someone else) has swapped.
In my opinion, others who've decided competently built cables are as good as all the exotica also have systems good enough that the cabling doesn't make a difference as well. Those that hear differences need to improve their sytems.
Hence, QED, my opinion is as valid as any of the others proclaimed w/o some sort of mutually agreed upon comparison set.
I've actually studied skin effect, for instance, at microwave and near infrared frequencies, oh 30 years ago or so at NRAO. Makes a big difference when transmitting extraterrestrial radiowaves through an .040" x .020" stainless steel waveguide to a mixer/detector in a helium filled dewer at near absolute zero. Coating the inside of the waveguide with electrodeposited copper to 3.5 skin depths at 150 GHZ turned out to be the best. Beat out thicker or thinner coatings, silver, gold, etc. And guess what? We used a screening multivariate design to minimize trials and treatment combinations and maximize success.
Go figure...
I'm right behind you, Terry... bye bye
So why should anyone believe there IS an audible difference in cables? Because THEY said so, according to JC, abrax..., curly, Monster Cable, MIT, etc. etc. what with all the hand waving and anti-multivariate design whining, not to mention add copy proclaiming all manner of irrelevant and un-demonstrated physical attributes and processes contributing to said difference...
OK.... I'm not convinced. My system is so good that changing cables has no effect on what I hear, even if I know that I (or someone else) has swapped.
In my opinion, others who've decided competently built cables are as good as all the exotica also have systems good enough that the cabling doesn't make a difference as well. Those that hear differences need to improve their sytems.
Hence, QED, my opinion is as valid as any of the others proclaimed w/o some sort of mutually agreed upon comparison set.
I've actually studied skin effect, for instance, at microwave and near infrared frequencies, oh 30 years ago or so at NRAO. Makes a big difference when transmitting extraterrestrial radiowaves through an .040" x .020" stainless steel waveguide to a mixer/detector in a helium filled dewer at near absolute zero. Coating the inside of the waveguide with electrodeposited copper to 3.5 skin depths at 150 GHZ turned out to be the best. Beat out thicker or thinner coatings, silver, gold, etc. And guess what? We used a screening multivariate design to minimize trials and treatment combinations and maximize success.
Go figure...
I'm right behind you, Terry... bye bye
Last edited:
Sorry but an article like that is just audiophile babble. Its stubborn old guys that are stuck on vinyl and analog that complain inaccurately to try and support their stubborness. Nothing new there.
The turth is not with thinking there is a "hazard of digital sound" from 30 years ago
Looks to me like solid engineering's being done there. I raised the issues of RF and grounding in passing a few posts back. This work shows its not sufficient to control merely the cables, one also needs to control various other factors in order to do a reliable DBT/ABX test.
Thanks for posting it up, will be interesting to see if the 'objectivists' just ignore it, dismiss it or ridicule it
- Status
- Not open for further replies.