Home |
Search |
Today's Posts |
#161
|
|||
|
|||
|
#162
|
|||
|
|||
Harry Lavo wrote:
wrote in message ... NO test can demonstrate that there is no difference. Any such test has one of two possible outcomes: 1) it can demonstrate that there IS a difference; 2) it can fail to determine whether there is or is not a difference. Of course, if you keep getting result #2, that should tell you something. Yes, either the test is wrong or you need more trials. :-) Basically, then, you are saying that if a test does not validate one's preconceptions it must be a bad test, or at least a good test wrongly implemented. Interesting approach, and no doubt a comforting one. Howard Ferstler |
#163
|
|||
|
|||
"Howard Ferstler" wrote in message
... Harry Lavo wrote: wrote in message ... NO test can demonstrate that there is no difference. Any such test has one of two possible outcomes: 1) it can demonstrate that there IS a difference; 2) it can fail to determine whether there is or is not a difference. Of course, if you keep getting result #2, that should tell you something. Yes, either the test is wrong or you need more trials. :-) Basically, then, you are saying that if a test does not validate one's preconceptions it must be a bad test, or at least a good test wrongly implemented. Interesting approach, and no doubt a comforting one. Howard Ferstler Howard, if you are going to reappear here, at least read the post in context and note the smiley. Obviously I was kidding following on the previous response. Do I have to be so literal as to spell it out for you? |
#164
|
|||
|
|||
Harry Lavo wrote:
"Howard Ferstler" wrote in message ... Harry Lavo wrote: wrote in message ... NO test can demonstrate that there is no difference. Any such test has one of two possible outcomes: 1) it can demonstrate that there IS a difference; 2) it can fail to determine whether there is or is not a difference. Of course, if you keep getting result #2, that should tell you something. Yes, either the test is wrong or you need more trials. :-) Basically, then, you are saying that if a test does not validate one's preconceptions it must be a bad test, or at least a good test wrongly implemented. Interesting approach, and no doubt a comforting one. Howard Ferstler Howard, if you are going to reappear here, at least read the post in context and note the smiley. Obviously I was kidding following on the previous response. Do I have to be so literal as to spell it out for you? The smiley notwithstanding, I tend to think that beneath it all you were serious about your contention. As for me reappearing, I really do read the threads once in a while, and I continue to marvel at all of those tempests in teapots. Rather than mix it up with the participants, I normally prefer to just roll my eyes and marvel at it all. In any case, I find it incredible that that people can go on and on about scientifically oriented testing. Note that I am not just referring to those who suspect such things as being misleading to audio truths, but also am referring to those who go on and on defending the procedure. Yep, both sides. The defenders of scientific protocol are nearly as wrong headed as those who attack it. If some people are not going to be convinced by a straightforward approach to audio testing and brass-tacks audio by now, it is a waste of time to step to the front of the room and try to convince them all over again. Audio is a hobby, after all, and trying to make it into hard science is just not going to work. Me? Well, I will continue to publish my little magazine articles and product reviews, and will occasionally post something here or on RAO (to be truthful, I have been posting WAY too much on RAO lately), but after seeing the monumental size of this series of threads, not to mention the series dealing with the LP vs the CD (and of course equally lengthy ones that have appeared in the past), I will leave it to people who are more enthusiastic about this "hobby" than I to engage in prolonged give and take. Howard Ferstler |
#165
|
|||
|
|||
"Howard Ferstler" wrote in message
... Harry Lavo wrote: "Howard Ferstler" wrote in message ... Harry Lavo wrote: wrote in message ... NO test can demonstrate that there is no difference. Any such test has one of two possible outcomes: 1) it can demonstrate that there IS a difference; 2) it can fail to determine whether there is or is not a difference. Of course, if you keep getting result #2, that should tell you something. Yes, either the test is wrong or you need more trials. :-) Basically, then, you are saying that if a test does not validate one's preconceptions it must be a bad test, or at least a good test wrongly implemented. Interesting approach, and no doubt a comforting one. Howard Ferstler Howard, if you are going to reappear here, at least read the post in context and note the smiley. Obviously I was kidding following on the previous response. Do I have to be so literal as to spell it out for you? The smiley notwithstanding, I tend to think that beneath it all you were serious about your contention. As for me reappearing, I really do read the threads once in a while, and I continue to marvel at all of those tempests in teapots. Rather than mix it up with the participants, I normally prefer to just roll my eyes and marvel at it all. Well, I'm happy to see you haven't given up your hobby of mind-reading. Even if you are wrong, as usual! :-) snip, irrelevant to comment here |
#166
|
|||
|
|||
Howard Ferstler wrote:
Harry Lavo wrote: (Snip irrelevant content) The defenders of scientific protocol are nearly as wrong headed as those who attack it. Well gosh Howard, maybe you should take this up with all those scientists who have been doing it so wrong all these years. If some people are not going to be convinced by a straightforward approach to audio testing and brass-tacks audio by now, it is a waste of time to step to the front of the room and try to convince them all over again. Howard, just a point of clarification, does straight forward mean sloppy and unreliable? What is "brass tacks," a new cone like tweak or code for your personal approach to audio or what? And don't people have to be convinced first before you convince them "all ove again?" Audio is a hobby, after all, and trying to make it into hard science is just not going to work. Well, I agree with that. but why such a devotion to DBTs if that is your position? DBTs are a tool of scientific investigation. Very useful in the hands of skilled researchers. Your claim that defenders of scientific protocols are wrong headed would make me think you are against it not only for the hobbyist but also for the pros. It's really hard to tell what you are saying now. (Snip personal anecdote.) Scott Wheeler |
#167
|
|||
|
|||
|
#168
|
|||
|
|||
Mark DeBellis wrote:
On 22 Jun 2005 03:06:23 GMT, wrote: analogies ... like visual ones, don't work. Roger Shepard argues the opposite ("It should not be surprising that some of the same underlying perceptual principles apply to more than one sensory domain"), giving numerous auditory/visual analogies, for example, between apparent visual motion and the perception of melodic lines (in his chapter on auditory stream segregation in Cook, ed., Music, Cognition, and Computerized Sound; the quote is from p. 127; pp. 21-35 likewise discuss general perceptual principles). Yes, but Roger Shepard understands the fields he is writing about. People who try to draw analogies between visual and aural perceptions without understanding either only wind up demonstrating how little they know. bob |
#169
|
|||
|
|||
|
#170
|
|||
|
|||
I would be curious to know if the research on perceptual coding (or
any other research) has turned up examples where two stimuli can't be discriminated from one another on close inspection, but where one stimulus is more fatiguing than the other over longer exposure. Mark |
#172
|
|||
|
|||
Mark DeBellis wrote:
OK, is this the idea. We want to prove that two sources provide the same information to the listener in the context of ordinary use. We appeal now to the following Principle: If two sources provide the same information in ordinary use, then in a test in which small, corresponding portions are compared, the listener will not be able to discriminate them. In particular, the type of discrimination task is the following: on any trial, two short excerpts are played, either both from the same source (a repetition) or from different sources, and the listener has to say if they're from the same or different sources, and he passes the test if he says same or different better than chance. Two observations. First of all, in a test like this there is plenty of information coming from either source that may not make it to the listener, such as the "temporally extended" properties I have mentioned. (Maybe that's obvious.) The thing is, the Principle says that that's OK because those properties have a certain dependence relation on the, as it were, "atomic" properties that get compared in such a test. The relation is, no difference in information presented unless there is a difference in atomic properties. In other words, information "supervenes" on atomic properties. It does so even if it does not *consist* exclusively in atomic properties (because, when I hear the Brahmsian style of a recorded performance, that is not any property of a short snippet). (To put the Principle better, information supervenes on *discernible* atomic properties.) I think you are right that Bob appeals to this principle, but I think I have a simple counterexample. Suppose a passage of music includes a gradual accelerando (quickening of tempo). If the accelerando is gradual enough, then it will be imperceptible in short segments, but clearly perceptible in the whole passage. Helen |
#173
|
|||
|
|||
Mark DeBellis wrote:
I would be curious to know if the research on perceptual coding (or any other research) has turned up examples where two stimuli can't be discriminated from one another on close inspection, but where one stimulus is more fatiguing than the other over longer exposure. It is highly unlikely that a researcher in psychoacoustics would consider this a matter worth investigating. Leaving aside the ill-defined term "fatiguing," there is neither evidence nor any plausible reason to believe that inaudible differences among codecs have any differential effect on the brain. bob |
#174
|
|||
|
|||
Helen Schmidt wrote:
I think you are right that Bob appeals to this principle, but I think I have a simple counterexample. Suppose a passage of music includes a gradual accelerando (quickening of tempo). If the accelerando is gradual enough, then it will be imperceptible in short segments, but clearly perceptible in the whole passage. The reason you won't be able to perceive it in the short segment is that it isn't happening in the short segment. I'll concede that we cannot hear things that are not there. (Will you?) But I am talking strictly about hearing differences between two separate presentations. Your example doesn't speak to that at all. bob |
#176
|
|||
|
|||
On 10 Jul 2005 17:24:00 GMT, "Helen Schmidt"
wrote: Mark DeBellis wrote: OK, is this the idea. We want to prove that two sources provide the same information to the listener in the context of ordinary use. We appeal now to the following Principle: If two sources provide the same information in ordinary use, then in a test in which small, corresponding portions are compared, the listener will not be able to discriminate them. In particular, the type of discrimination task is the following: on any trial, two short excerpts are played, either both from the same source (a repetition) or from different sources, and the listener has to say if they're from the same or different sources, and he passes the test if he says same or different better than chance. Two observations. First of all, in a test like this there is plenty of information coming from either source that may not make it to the listener, such as the "temporally extended" properties I have mentioned. (Maybe that's obvious.) The thing is, the Principle says that that's OK because those properties have a certain dependence relation on the, as it were, "atomic" properties that get compared in such a test. The relation is, no difference in information presented unless there is a difference in atomic properties. In other words, information "supervenes" on atomic properties. It does so even if it does not *consist* exclusively in atomic properties (because, when I hear the Brahmsian style of a recorded performance, that is not any property of a short snippet). (To put the Principle better, information supervenes on *discernible* atomic properties.) I think you are right that Bob appeals to this principle, but I think I have a simple counterexample. Suppose a passage of music includes a gradual accelerando (quickening of tempo). If the accelerando is gradual enough, then it will be imperceptible in short segments, but clearly perceptible in the whole passage. Agreed, and in this particular case, it's necessary to listen to the entire piece. That would of course make such a musical piece a very poor choice for doing comparisons between two pieces of equipment. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#177
|
|||
|
|||
Mark DeBellis wrote:
wrote: Mark DeBellis wrote: I would be curious to know if the research on perceptual coding (or any other research) has turned up examples where two stimuli can't be discriminated from one another on close inspection, but where one stimulus is more fatiguing than the other over longer exposure. It is highly unlikely that a researcher in psychoacoustics would consider this a matter worth investigating. Leaving aside the ill-defined term "fatiguing," Ah, thanks for the reply. The notion of "fatiguing" could be given an operational definition by, for example, allowing the subject to keep listening to the song played over and over until he/she decides to switch it off. There are rules against using cruelty in scientific experiments on human subjects. there is neither evidence nor any plausible reason to believe that inaudible differences among codecs have any differential effect on the brain. OK, so the question is, are there stimuli that on close inspection are indistinguishable from one another but would get rated differently according to the above way of defining "fatiguing"? And from what you're saying it would seem to follow that no one knows. There is no evidence that the answer is yes, but then again no one has investigated it, so we shouldn't conclude too much from the absence of such evidence. Or is there reason to think that there could *not* be such stimuli--that stimuli that are indistinguishable on close inspection would *have* to get similar ratings for fatigue as described above, as a matter of psychological fact? If so, what does that reason look like? What's the principle behind it? Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. IOW, if you can't hear it in a listening test, we're on pretty safe ground in assuming that it can't excite the aural nerve, and therefore that your brain cannot react to it in any way. That's why psychoacousticians use these listening tests--because they've proven sensitive enough to do the job. bob |
#178
|
|||
|
|||
wrote:
Mark DeBellis wrote: wrote: Mark DeBellis wrote: I would be curious to know if the research on perceptual coding (or any other research) has turned up examples where two stimuli can't be discriminated from one another on close inspection, but where one stimulus is more fatiguing than the other over longer exposure. It is highly unlikely that a researcher in psychoacoustics would consider this a matter worth investigating. Leaving aside the ill-defined term "fatiguing," Ah, thanks for the reply. The notion of "fatiguing" could be given an operational definition by, for example, allowing the subject to keep listening to the song played over and over until he/she decides to switch it off. There are rules against using cruelty in scientific experiments on human subjects. there is neither evidence nor any plausible reason to believe that inaudible differences among codecs have any differential effect on the brain. OK, so the question is, are there stimuli that on close inspection are indistinguishable from one another but would get rated differently according to the above way of defining "fatiguing"? And from what you're saying it would seem to follow that no one knows. There is no evidence that the answer is yes, but then again no one has investigated it, so we shouldn't conclude too much from the absence of such evidence. Or is there reason to think that there could *not* be such stimuli--that stimuli that are indistinguishable on close inspection would *have* to get similar ratings for fatigue as described above, as a matter of psychological fact? If so, what does that reason look like? What's the principle behind it? Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. OK, that I get, but ... We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. IOW, if you can't hear it in a listening test, we're on pretty safe ground in assuming that it can't excite the aural nerve, and therefore that your brain cannot react to it in any way. .... how do we get from "a sound that is too quiet to be heard" to "a difference that is too small to detect"? Are you saying that whenever two sounds are indistinguishable then the input to the auditory nerve is the same? Not sure I get the relation between "if you can't hear it" in the sense that you don't hear a sound and the sense that you don't hear (detect) a difference... not sure that they're the same thing really. Mark |
#179
|
|||
|
|||
|
#180
|
|||
|
|||
wrote in message ...
Mark DeBellis wrote: wrote: Mark DeBellis wrote: I would be curious to know if the research on perceptual coding (or any other research) has turned up examples where two stimuli can't be discriminated from one another on close inspection, but where one stimulus is more fatiguing than the other over longer exposure. It is highly unlikely that a researcher in psychoacoustics would consider this a matter worth investigating. Leaving aside the ill-defined term "fatiguing," Ah, thanks for the reply. The notion of "fatiguing" could be given an operational definition by, for example, allowing the subject to keep listening to the song played over and over until he/she decides to switch it off. There are rules against using cruelty in scientific experiments on human subjects. there is neither evidence nor any plausible reason to believe that inaudible differences among codecs have any differential effect on the brain. OK, so the question is, are there stimuli that on close inspection are indistinguishable from one another but would get rated differently according to the above way of defining "fatiguing"? And from what you're saying it would seem to follow that no one knows. There is no evidence that the answer is yes, but then again no one has investigated it, so we shouldn't conclude too much from the absence of such evidence. Or is there reason to think that there could *not* be such stimuli--that stimuli that are indistinguishable on close inspection would *have* to get similar ratings for fatigue as described above, as a matter of psychological fact? If so, what does that reason look like? What's the principle behind it? Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. IOW, if you can't hear it in a listening test, we're on pretty safe ground in assuming that it can't excite the aural nerve, and therefore that your brain cannot react to it in any way. That's why psychoacousticians use these listening tests--because they've proven sensitive enough to do the job. That's the conventional wisdom, based on older research, I believe. More recent research suggests the brain actively changes the sensitivity and more importantly the "focus" of what the ears are to listen for and process depending on what type of sound is involved, even changing the physical configuration of the sensors. And in the process, the work has apparently reemphasized the importance of distortion and transient perception. In other words, the ear/brain complex listens "differently" to music, to telephone voices, to threatening sounds in the woods, to traffic noise, etc. Don't ask me for sources...I've picked this up second-hand via NPR, Psych Today, the NYT, etc. But it's out there somewhere if one wants to do the research. And BTW, when Bob says "it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve." he is talking about sound per se (usually white noise), not music. These tests are less sensitive for music. And as now seems possible, it may be because the brain is controlling the ear to focus on different things when listening to music. |
#181
|
|||
|
|||
wrote:
Mark DeBellis wrote: wrote: Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. I take it that the above is not true of subjects who are tone deaf through some brain pathology, i.e., who cannot discriminate pitches normally, but who have nothing wrong with their ears? Mark |
#182
|
|||
|
|||
Mark DeBellis wrote:
... how do we get from "a sound that is too quiet to be heard" to "a difference that is too small to detect"? Are you saying that whenever two sounds are indistinguishable then the input to the auditory nerve is the same? Not sure I get the relation between "if you can't hear it" in the sense that you don't hear a sound and the sense that you don't hear (detect) a difference... not sure that they're the same thing really. You're pushing the bounds of my "expertise," so I hope someone who knows this stuff better than I will chime in, but basically there's no difference because in both cases you're talking about detecting a tiny change in air pressure--which, as it turns out, is the only thing the ear can do. bob |
#183
|
|||
|
|||
In article ,
"Mark DeBellis" wrote: wrote: Mark DeBellis wrote: wrote: Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. I take it that the above is not true of subjects who are tone deaf through some brain pathology, i.e., who cannot discriminate pitches normally, but who have nothing wrong with their ears? Mark Just a note here, and I'm sure that you're already aware of this: true tone deafness is VERY rare. There are indeed people for whom you can play "Mary Had a Little Lamb", then play it again with half of the notes being "wrong" and they can't tell the difference. See the work of Peretz in Montreal and others. But this is so rare that few of us here have ever met such a person. The vast majority of people who call themselves or others "tone deaf" because they don't seem to be able to sing in tune, etc. are simply untrained in the ear/voice/brain connection, and aren't really tone deaf at all. It's a very over used term, I'm afraid. |
#184
|
|||
|
|||
On 12 Jul 2005 23:41:58 GMT, "Mark DeBellis"
wrote: wrote: Mark DeBellis wrote: wrote: Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. I take it that the above is not true of subjects who are tone deaf through some brain pathology, i.e., who cannot discriminate pitches normally, but who have nothing wrong with their ears? If we are distinguishing two pieces of audio equipment due to pitch difference, then something is *seriously* broken! :-) -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#186
|
|||
|
|||
Stewart Pinkerton wrote:
If we are distinguishing two pieces of audio equipment due to pitch difference, then something is *seriously* broken! :-) Just maybe to clarify ... although at this point in the discussion I remain skeptical, from a theoretical point of view, about how much discrimination tests actually tell us ... .... as a practical matter, if I knew that people could not tell the difference between two given components in blind testing, then I wouldn't really consider spending (say) $500 more on one of them. I would be very skeptical about any claim that one was better. But I might spend (say) $8 more on the SACD over the CD, even if I'd heard that any difference was very hard to discern (and even given that I'd found in direct comparison that the differences were much smaller than I'd expected them to be). $8 is not a lot to bet that the received wisdom is wrong. In other words, I am skeptical both about the claim that one thing is better than another and about the claim that it is known that one is not better than the other. I also think I am not totally irrational about how to translate this into dollars and cents, although perhaps others would not agree ... I also think that if you told me that an inexpensive, well-built power cord is just as good as the $1000 power cords being sold, I would believe you, but my confidence in what you told me would be based not so much on the outcome of listening tests as on your general engineering and scientific knowledge of how the relevant devices work and how the output signal could, or could not, possibly be affected by a difference between the cords. Theory counts for a lot! Again, this may be obvious to everyone else already. Mark |
#187
|
|||
|
|||
|
#188
|
|||
|
|||
Stewart Pinkerton wrote:
On 12 Jul 2005 23:41:58 GMT, "Mark DeBellis" wrote: wrote: Mark DeBellis wrote: wrote: Very simply, we can measure signals passing through the aural (otic?) nerve to the brain. And we can, and have, figured out the quietest sounds that will trigger a signal to the brain. I presume you will accept the fact that if something is too quiet to excite the aural nerve, then the brain cannot react to it in any way. We've also used listening tests to determine the quietest sounds that humans can detect. And it turns out that the quietest sounds we can detect in listening tests are pretty close to the quietest sounds that can excite the aural nerve. I take it that the above is not true of subjects who are tone deaf through some brain pathology, i.e., who cannot discriminate pitches normally, but who have nothing wrong with their ears? If we are distinguishing two pieces of audio equipment due to pitch difference, then something is *seriously* broken! :-) Fair enough, but I guess what I am trying to say is that, quite possibly, the objectivist shouldn't put all the weight of his/her argument on the discrimination test per se. By itself, I don't know how much it proves. If there are other, general theoretical reasons to think there couldn't be a detectable difference in sound between two sources, or if any difference is constrained to be one of a very limited range, then that bolsters the argument. Again, this may be obvious to everyone else, but I'm just seeing it now. (The main example I am envisioning here is SACD vs. CD, not comparing two cables say, where the possible physical variation would, I assume, be very highly constrained.) Anyway, I mean the point about tone-deafness as a general methodological point about the relation between the following: 1. When one stimulus evokes a just-barely-measurable neural signal, and another evokes no signal at all, the subject can discriminate the two stimuli from one another, and 2. For *any* two stimuli (not just very quiet ones), if the subject cannot discriminate them, then the neural signals cannot be measurably different. (I take it that Bob was arguing for 2 on the basis of 1.) If 2 is supposed to follow from 1, then 1 had better not be true for any subjects who cannot discriminate pitches due to brain injury. If it were, this would show that the inference from 1 to 2 is not valid, since for such a person 1 would be true but 2 false. Mark |
#189
|
|||
|
|||
Mark DeBellis wrote:
wrote: Mark DeBellis wrote: ... how do we get from "a sound that is too quiet to be heard" to "a difference that is too small to detect"? Are you saying that whenever two sounds are indistinguishable then the input to the auditory nerve is the same? Not sure I get the relation between "if you can't hear it" in the sense that you don't hear a sound and the sense that you don't hear (detect) a difference... not sure that they're the same thing really. You're pushing the bounds of my "expertise," so I hope someone who knows this stuff better than I will chime in, but basically there's no difference because in both cases you're talking about detecting a tiny change in air pressure--which, as it turns out, is the only thing the ear can do. Well, this is certainly not in my expertise either, but honestly I don't see how the result transfers to arbitrary contexts. In one case we're talking about a tiny difference in isolation; Not really. There's always background noise. So you're really comparing background-noise to background-noise-plus-tiny signal. that will be discriminated. In the general case there are two complex sounds that differ by at least that tiny amount; does it follow that, in that more complicated context, the listener will be able to discriminate between them? I just don't see what compels the conclusion that, if the tiny amount can be detected in isolation, then it can also be detected in the more complicated context. Not what I said at all, and irrelevant to the point we were discussing. bob |
#191
|
|||
|
|||
OK, so to come back to what is basically the original question.
Suppose Karl is comparing SACD and CD of the same recording, and he fails a discrimination test. This means he takes a quick-switch ABX test and fails to discriminate at a significant level one source from the other, as same or different. (Note: this is not the kind of test I reported failing in the SACD/CD thread; what I failed was an identification test, not a discrimination test.) The skeptic claims: for all we know from the result of this test, there could be a perceptual difference between the two sources in normal listening (as opposed to testing) situations. For all we know, in listening to 5-minute stretches, say, Karl derives greater satisfaction or pleasure from one source than he does from the other. Ah, but the ABX-proponent says, then let him listen first to one stretch, then the other, and if he derives greater pleasure from one than the other, then he *will* be able to discriminate them. The skeptic replies, not necessarily. He will be able to do this only if, say, he is good at *comparing* the pleasure he derives from one passage with the pleasure that he derives from the other. And he may not be good at this if he is not able to retain an accurate representation in memory of the pleasure derived from the first while he listens to the second, or if he is just not very good at comparing two memory traces. In other words, to say he *must* be able to discriminate in such a case is to impute to him a greater power of introspection than he may in fact have. OK, then the ABXer's response is that the skeptic's notion that two experiences can differ in satisfaction or pleasure derived, if they cannot be reliably discriminated as such, is empirically meaningless. The skeptic replies: not necessarily. For example, if during listening Karl is asked to rate his satisfaction on a scale from 1 to 10, and if the ratings for one source are significantly different from those of the other, that would be an empirical difference. (And it wouldn't follow from the fact that the ratings are significantly different that Karl would be able to pass a *discrimination* test. Differential response, in one situation, does not necessarily imply an ability to *compare* reliably, in a much different situation.) I think this idea of ratings is basically the same in spirit as Harry's suggestion of "monadic" tests. Mark |
#192
|
|||
|
|||
On 14 Jul 2005 23:20:14 GMT, "Mark DeBellis"
wrote: 1. When one stimulus evokes a just-barely-measurable neural signal, and another evokes no signal at all, the subject can discriminate the two stimuli from one another, and 2. For *any* two stimuli (not just very quiet ones), if the subject cannot discriminate them, then the neural signals cannot be measurably different. (I take it that Bob was arguing for 2 on the basis of 1.) If 2 is supposed to follow from 1, then 1 had better not be true for any subjects who cannot discriminate pitches due to brain injury. If it were, this would show that the inference from 1 to 2 is not valid, since for such a person 1 would be true but 2 false. One thing should be tidied up here. The effect known as masking does make detectable partial loudness intervals larger as the general SPL goes up, i.e. we can not hear sounds which are more than about 45dB below the average sound level. Perceptual codecs such as MP3 rely on this feature of human hearing. That's one reason why vinylphiles claim not to be troubled by surface noise. -- Stewart Pinkerton | Music is Art - Audio is Engineering |
#193
|
|||
|
|||
Mark DeBellis wrote:
OK, so to come back to what is basically the original question. Suppose Karl is comparing SACD and CD of the same recording, and he fails a discrimination test. This means he takes a quick-switch ABX test and fails to discriminate at a significant level one source from the other, as same or different. (Note: this is not the kind of test I reported failing in the SACD/CD thread; what I failed was an identification test, not a discrimination test.) The skeptic claims: for all we know from the result of this test, there could be a perceptual difference between the two sources in normal listening (as opposed to testing) situations. For all we know, in listening to 5-minute stretches, say, Karl derives greater satisfaction or pleasure from one source than he does from the other. Ah, but the ABX-proponent says, then let him listen first to one stretch, then the other, and if he derives greater pleasure from one than the other, then he *will* be able to discriminate them. The skeptic replies, not necessarily. He will be able to do this only if, say, he is good at *comparing* the pleasure he derives from one passage with the pleasure that he derives from the other. And he may not be good at this if he is not able to retain an accurate representation in memory of the pleasure derived from the first while he listens to the second, or if he is just not very good at comparing two memory traces. In other words, to say he *must* be able to discriminate in such a case is to impute to him a greater power of introspection than he may in fact have. Except that, if he's unable to do this blind, then he is unable to do it sighted. If he's no good at comparing the pleasure he derives from each, then how did he arrive at the original judgment that SACD sounds better than CD? (And if he didn't start from this, why even do the blind test?) And assuming he did somehow arrive at the original judgment that SACD sounds better than CD, why can he not duplicate that comparison blind, multiple times, to see if he arrives at the same judgment consistently? I'll tell you why--because he's afraid of the result. OK, then the ABXer's response is that the skeptic's notion that two experiences can differ in satisfaction or pleasure derived, if they cannot be reliably discriminated as such, is empirically meaningless. The skeptic replies: not necessarily. For example, if during listening Karl is asked to rate his satisfaction on a scale from 1 to 10, and if the ratings for one source are significantly different from those of the other, that would be an empirical difference. "If" isn't empirical. All the empirical evidence suggests that a test like this would fail, for reasons you've been given over and over and over again. Give it a rest. (And it wouldn't follow from the fact that the ratings are significantly different that Karl would be able to pass a *discrimination* test. Differential response, in one situation, does not necessarily imply an ability to *compare* reliably, in a much different situation.) Opinion stated as fact. There is no evidence that "discrimination" in hearing and "comparing" in hearing are different. You are merely playing at semantics, without a shred of evidence to back you up. I think this idea of ratings is basically the same in spirit as Harry's suggestion of "monadic" tests. Yes, it does sound like you have bought into Harry's anti-scientific worldview. I suggest you endeavor to talk to a real scientist for a change. bob |
#194
|
|||
|
|||
Mark DeBellis wrote:
snip The skeptic claims: for all we know from the result of this test, there could be a perceptual difference between the two sources in normal listening (as opposed to testing) situations. For all we know, in listening to 5-minute stretches, say, Karl derives greater satisfaction or pleasure from one source than he does from the other. To postulate this requires stipulating that Karl did, indeed, make a distinction. Ah, but the ABX-proponent says, then let him listen first to one stretch, then the other, and if he derives greater pleasure from one than the other, then he *will* be able to discriminate them. That is *implicit*. You cannot enjoy one more than the other without being able to discriminate between them. There would be *no* basis. The skeptic replies, not necessarily. He will be able to do this only if, say, he is good at *comparing* the pleasure he derives from one passage with the pleasure that he derives from the other. If not, then he cannot, by definition, know that he enjoyed one more than another. And he may not be good at this if he is not able to retain an accurate representation in memory of the pleasure derived from the first while he listens to the second, or if he is just not very good at comparing two memory traces. In other words, to say he *must* be able to discriminate in such a case is to impute to him a greater power of introspection than he may in fact have. Nonsense. You have already *stipulated* that he enjoyed one more than another. That very fact *defines* extant discrimination. OK, then the ABXer's response is that the skeptic's notion that two experiences can differ in satisfaction or pleasure derived, if they cannot be reliably discriminated as such, is empirically meaningless. No, rather, that which is not repeatable is not considered valid empirical data. The skeptic replies: not necessarily. For example, if during listening Karl is asked to rate his satisfaction on a scale from 1 to 10, and if the ratings for one source are significantly different from those of the other, that would be an empirical difference. And if these results were not repeatable, then they would be empirically meaningless, within the context of a verifiable difference (sonically, emotionally, musically, whatever). Further, if the tests were not relatively contemporaneous, Karl's emotional state would be as likely, if not more so, to be responsible for a disparate reaction (between A and B for e.g.) than would be an actual audible difference. Certainly it would introduce sufficient doubt as to invalidate the test. (And it wouldn't follow from the fact that the ratings are significantly different that Karl would be able to pass a *discrimination* test. He just *did* 'pass' a discrimination test, by definition. That he didn't say "A is better than B" is irrelevant. If he states that "A engenders a pleasure level of 9 and B engenders a pleasure level of 2" he has clearly discriminated between the systems. Differential response, in one situation, does not necessarily imply an ability to *compare* reliably, in a much different situation.) Reproducibility is the hallmark of *valid* data, irrespective of context. If, in the context the orginal observation was made, the observation cannot be repeated, it must be considered either an anomaly, or simply erroneous. There are no other interpretations, although there are a myriad of possible underlying causes. I think this idea of ratings is basically the same in spirit as Harry's suggestion of "monadic" tests. Mark |
#195
|
|||
|
|||
wrote in message ...
Yes, it does sound like you have bought into Harry's anti-scientific worldview. I suggest you endeavor to talk to a real scientist for a change. Nothing I have said in this or any other group is "unscientific". It does not support ABX without questioning some of the basic premises. That simple. If you are truly a scientist, you think about these things objectively, at least. If you have become dogmatic and accept it as an article of faith, then questioning it becomes tauntamont to heresy. I leave readers to decide which is operational here. |
#196
|
|||
|
|||
"Harry Lavo" wrote in message
... wrote in message ... Yes, it does sound like you have bought into Harry's anti-scientific worldview. I suggest you endeavor to talk to a real scientist for a change. Nothing I have said in this or any other group is "unscientific". It does not support ABX without questioning some of the basic premises. That simple. If you are truly a scientist, you think about these things objectively, at least. If you have become dogmatic and accept it as an article of faith, then questioning it becomes tauntamont to heresy. I leave readers to decide which is operational here. I'm forever interested in how those of Harry's persuasion would go about determining if there is an audible difference between 2 devices. The question is always the same. Can you hear a difference between 2 presentations? You can listen as long as you wish; you can ask other opinions; you can switch between them slowly or quickly; you can take notes; you can change the source material, etc., etc. The only things you cannot do is peek or make measurements. IOW, you have to make your choices on the basis of what you hear. To my way of thinking, ABX seems to be one of the more sensitive methods of making such a determination. But if you don't, suggest something else. I'm certainly open to suggestions. One last point: DBT seems to be the gold standard for evaluating most everything else. Why should high fidelity audio be an exception? Norm Strong |
#197
|
|||
|
|||
wrote in message
... "Harry Lavo" wrote in message ... wrote in message ... Yes, it does sound like you have bought into Harry's anti-scientific worldview. I suggest you endeavor to talk to a real scientist for a change. Nothing I have said in this or any other group is "unscientific". It does not support ABX without questioning some of the basic premises. That simple. If you are truly a scientist, you think about these things objectively, at least. If you have become dogmatic and accept it as an article of faith, then questioning it becomes tauntamont to heresy. I leave readers to decide which is operational here. I'm forever interested in how those of Harry's persuasion would go about determining if there is an audible difference between 2 devices. The question is always the same. Can you hear a difference between 2 presentations? You can listen as long as you wish; you can ask other opinions; you can switch between them slowly or quickly; you can take notes; you can change the source material, etc., etc. The only things you cannot do is peek or make measurements. IOW, you have to make your choices on the basis of what you hear. To my way of thinking, ABX seems to be one of the more sensitive methods of making such a determination. But if you don't, suggest something else. I'm certainly open to suggestions. One last point: DBT seems to be the gold standard for evaluating most everything else. Why should high fidelity audio be an exception? Norman, I've explained in detail here and on other forums exactly how a blind, monadic test would be used to detect and define (statistically) a difference, if it existed. Then ABX could be tested to see if it revealed those differences. |
#198
|
|||
|
|||
Many of these questions have been dicussed before. Using an ABX is NOT
equivalent to DBT as is used in drug testing. The ABX machine itself is a limitation, especially if interconnect is neing evaluated. How good are the contacts? How good is the internal mating? The problem in any audio evaluation is that we MUST rely on audio memory. True DBT (such as drug DBT) has no such limitation. It is IMPOSSIBLE to listen simultaneously to two different components. One is always listening to one or the other, while holding in one's memory the sound of the other. I have heard differences between various interconnects. I performed medium-term listening in the dark, so that my sense of hearing could hone in on the matter at hand. Evaluative listening undertaken with the lights on is not as precise, beacause the brain is occupied with vision, which is very demanding on the brain. 'Going dark' is VERY revealing. Those who SWEAR they cannot hear differences in cables should try it! wrote: "Harry Lavo" wrote in message ... wrote in message ... Yes, it does sound like you have bought into Harry's anti-scientific worldview. I suggest you endeavor to talk to a real scientist for a change. Nothing I have said in this or any other group is "unscientific". It does not support ABX without questioning some of the basic premises. That simple. If you are truly a scientist, you think about these things objectively, at least. If you have become dogmatic and accept it as an article of faith, then questioning it becomes tauntamont to heresy. I leave readers to decide which is operational here. I'm forever interested in how those of Harry's persuasion would go about determining if there is an audible difference between 2 devices. The question is always the same. Can you hear a difference between 2 presentations? You can listen as long as you wish; you can ask other opinions; you can switch between them slowly or quickly; you can take notes; you can change the source material, etc., etc. The only things you cannot do is peek or make measurements. IOW, you have to make your choices on the basis of what you hear. To my way of thinking, ABX seems to be one of the more sensitive methods of making such a determination. But if you don't, suggest something else. I'm certainly open to suggestions. One last point: DBT seems to be the gold standard for evaluating most everything else. Why should high fidelity audio be an exception? Norm Strong |
#199
|
|||
|
|||
|
#200
|
|||
|
|||
|
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
common mode rejection vs. crosstalk | Pro Audio | |||
Topic Police | Pro Audio |