Home |
Search |
Today's Posts |
#201
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
|
#202
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 20, 6:21*pm, wrote:
On May 20, 7:06*pm, wrote: What "critical aspects" were known by either the subject or adminstrator that would prevent Sonova's test from being considered double blind? The test administrator's knowledge of where the switch was. That's all it would take. Bingo! That said, I think people are nitpicking this whole question too much. There are dozens of ways to screw up a DBT (in both directions). How Sonnova did it, if he did it, we'll never know because none of us were there. What we do know is that people who perform these tests more carefully and rigorously then he did, including the developers of MP3, invariably get a very different result than he did. Now maybe his test is right and everyone else's is wrong. But I doubt it. I agree totally with this last bit as well. It is quite hard to do science well. Even scientists who have spend decades doing science still muck it up quite frequently. That's what peer review is for. It doesn't make it any easier if you go in not knowing just what kinds of mistakes to avoid. |
#203
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
|
#205
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
"Steven Sullivan" wrote in message ... Sonnova wrote: On Sun, 17 May 2009 12:13:52 -0700, wrote (in article ): On May 17, 11:30?am, wrote: On May 16, 9:56?am, Steven Sullivan wrote: wrote: Are double blind tests only as good as many are claiming here on RAHE when they wrought the desired result? No, but a remarkable result requires remarkable evidence, don't you agree? It seems to me that some are calling results remarkable because they personally have some sort of prejudicial issue with them. I don't see anything remarkable about Sonova's results. Who is the arbitrator of remarkable? But, no, I don't believe in a moving goal post for evidence based on prejudice. I think the standards should be, well, standard. Whenever you get an outlier result?and Sonnova's result is certainly an outlier?the first thing you need to do is take a close look at the method, and probably repeat the experiment. Any good scientist will tell you that when you get an unexpected result, you probably screwed up the experiment. bob Why should such a result be unexpected? Lossy compression schemes THROW INFORMATION AWAY based on somebody's definition of what's not important. THis is a misleading use of language. "Somebody" in this case happens to be the body of psychoacoustic data concerning masking and thresholds. It's not arbitrary. It's also not automatic, Steven. Folks worked for years to get compression schemes that were even acceptable to a large majority of trained listeners. There is no guarantee that an especially acute listener and one who has trained himself in the artifacts to hear them even when most people cannot. Yes, there is no guarantee. But it's not predicted to be common. Therefore it's expected to be unusual. Therefore really good evidence is required. I haven't seen a lick of evidence that Sonnova has *trained* himself or is an especially acute listener. He seems rather uninformed about mp3s generally, in fact. Once one throws info away, its gone - forever. Why is the fact that some people can detect, in double-blind tests, the results of throwing this info away, or as by-products of the process, an unexpected result? Because YOU don't hear it? No one is saying it's impossible to detect. That is not the same as saying it is routeinly detectable under all circumstances. Certainly for high-bitrate mp3, a rigorous DBT method is required to demonstrate that detection has occurred, and conditions need to be specified. It would not be unusual for many to detect a difference if a 'killer' sample is used, or a poor encoder is used, or if the conversion of mp3s to WAVs for testing produced clipping, for example. Or that the listener is well trained, has very fine equipment, and listen's mostly to the kinds of music (as in Sonnova's case) where artifacts are most likely to be heard. Who says Sonnova's music's the type where artifacts are most likely to be heard? Do you know what types of music are? Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. -- -S We have it in our power to begin the world over again - Thomas Paine |
#206
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Sonnova" wrote in message
On Wed, 20 May 2009 09:35:36 -0700, Arny Krueger wrote (in article ): "Sonnova" wrote in message Others tell me that MP3 works. Compared to the LP format, it works brilliantly. Your prejudice is showing again, Arny. Prejudice? So you're saying that the 35+ years that I spent listening to little but LPs is not a valid way to evaluate the sound quality of the LP format? Of course, LP quality varies all over the map, but I have LPs that sound much more appealing and less fatiguing than do any MP3 I've ever heard. That's a choice you get to make, but you don't have to impugn my motives and attack my character to exercise it. My ears tell me that it doesn't. So you're saying that your brain has nothing to do with it? Trying to be funny? Making a point. The brain has a lot to do with what you hear. I didn't listen to those who's tests were, in your words, "properly conducted" and found that they were wrong. Given how easy they are to do, this speaks to a closed mind. Not closed. Just not interested in revisiting a subject that I have already investigated and made up my mind about Absolutely symptomatic of a closed mind. BTW, I still listen to LPs. |
#207
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
wrote in message ...
On May 20, 11:57 am, Ed Seedhouse wrote: On May 20, 8:33 am, Sonnova wrote: On Tue, 19 May 2009 16:49:37 -0700, Ed Seedhouse wrote Well, yes you did, sorry. There are well known mechanisms that allow that which your experimental protocol didn't rule out. List them, please. I have already given some applicable search terms in my previous posts, which you don't seem to have checked into. The mechanisms are well known and widely documented. If you don't want to research them that's your right. It is, however, rather worrying that you don't already know about them yet are making claims to have done a "double blind" test. "Noun 1. double-blind experiment - an experimental procedure in which neither the subjects of the experiment nor the persons administering the experiment know the critical aspects of the experiment; "a double- blind procedure is used to guard against both experimenter bias and placebo effects" " What "critical aspects" were known by either the subject or adminstrator that would prevent Sonova's test from being considered double blind? The results, obviously. |
#208
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 21, 7:56*am, Sonnova wrote:
The test administrator's knowledge of where the switch was. That's all it would take. There is no "switch". Certainly there was, given that your description of the process is accurate. The CD player was set to "shuffle" among the cuts. And there it is. All she knew was what cut NUMBER was playing, There's the knowledge of the switch. If a track number was repeated then she also had knowledge of your previous reaction to that particular track, which could have influenced you when the track repeated since she would have expectations based on your previous reaction which could have affected her. She might easily have given a facial or body language clue to you that might easily have influenced you. Even if you did not see her you could still, given the description she gave, have heard any slight reactions she might have involuntarily made, such as a slight cougn, or movement in her chair. Given that you could not observe her reactions, that's still single blind because she could observe your reactions. If you could see or hear her reactions to the track number then it would just be a busted test. |
#209
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Steven Sullivan" wrote in message
... Arny Krueger wrote: "Harry Lavo" wrote in message It's also not automatic, Steven. Folks worked for years to get compression schemes that were even acceptable to a large majority of trained listeners. "large majority of trained listeners" - an undefined group if there ever was one. I have no idea who he is even talking about. There aren't that many trained listeners for mp3 artifacts. JJ and others have indicated that listeners are first screened for basic accuity, and then must be specifically trained to recognize and differentiate the artifact under question before a valid test can be done. That's what a trained listener is/was. In the early days, this led to more tests getting confirmed as different than did tests that indicated "no difference". Mostly irrelevant because it doesn't take very fine equipment to hear most artifacts. In fact, most LAME codec tweaking has been done using computer speakers and headphones. So perhaps they missed the kind that Sonnova picked up? How to you test "air" on a headphone? When phase is largely taken out of the picture? |
#210
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Steven Sullivan" wrote in message
... Harry Lavo wrote: "Steven Sullivan" wrote in message ... Sonnova wrote: On Sun, 17 May 2009 12:13:52 -0700, wrote (in article ): On May 17, 11:30?am, wrote: On May 16, 9:56?am, Steven Sullivan wrote: wrote: Are double blind tests only as good as many are claiming here on RAHE when they wrought the desired result? No, but a remarkable result requires remarkable evidence, don't you agree? It seems to me that some are calling results remarkable because they personally have some sort of prejudicial issue with them. I don't see anything remarkable about Sonova's results. Who is the arbitrator of remarkable? But, no, I don't believe in a moving goal post for evidence based on prejudice. I think the standards should be, well, standard. Whenever you get an outlier result?and Sonnova's result is certainly an outlier?the first thing you need to do is take a close look at the method, and probably repeat the experiment. Any good scientist will tell you that when you get an unexpected result, you probably screwed up the experiment. bob Why should such a result be unexpected? Lossy compression schemes THROW INFORMATION AWAY based on somebody's definition of what's not important. THis is a misleading use of language. "Somebody" in this case happens to be the body of psychoacoustic data concerning masking and thresholds. It's not arbitrary. It's also not automatic, Steven. Folks worked for years to get compression schemes that were even acceptable to a large majority of trained listeners. There is no guarantee that an especially acute listener and one who has trained himself in the artifacts to hear them even when most people cannot. Yes, there is no guarantee. But it's not predicted to be common. Therefore it's expected to be unusual. Therefore really good evidence is required. I haven't seen a lick of evidence that Sonnova has *trained* himself or is an especially acute listener. He seems rather uninformed about mp3s generally, in fact. That's his claim; you can be skeptical but that doesn't make you correct. Once one throws info away, its gone - forever. Why is the fact that some people can detect, in double-blind tests, the results of throwing this info away, or as by-products of the process, an unexpected result? Because YOU don't hear it? No one is saying it's impossible to detect. That is not the same as saying it is routeinly detectable under all circumstances. Certainly for high-bitrate mp3, a rigorous DBT method is required to demonstrate that detection has occurred, and conditions need to be specified. It would not be unusual for many to detect a difference if a 'killer' sample is used, or a poor encoder is used, or if the conversion of mp3s to WAVs for testing produced clipping, for example. Or that the listener is well trained, has very fine equipment, and listen's mostly to the kinds of music (as in Sonnova's case) where artifacts are most likely to be heard. Who says Sonnova's music's the type where artifacts are most likely to be heard? Do you know what types of music are? Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? Well, I listened myself for as long as I could stand it on my "very fine" equipment. So I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of antecdotal evidence around. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. I'm not arguing that headphones can't demonstrate certain distortions...just that they have never been tested to show that they can demonstrate ALL distortions. It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. If you are a scientist. Which you are not. |
#211
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On Thu, 21 May 2009 11:56:44 -0700, Ed Seedhouse wrote
(in article ): On May 21, 7:56*am, Sonnova wrote: The test administrator's knowledge of where the switch was. That's all it would take. There is no "switch". Certainly there was, given that your description of the process is accurate. The CD player was set to "shuffle" among the cuts. And there it is. All she knew was what cut NUMBER was playing, There's the knowledge of the switch. If a track number was repeated then she also had knowledge of your previous reaction to that particular track, which could have influenced you when the track repeated since she would have expectations based on your previous reaction which could have affected her. She might easily have given a facial or body language clue to you that might easily have influenced you. Even if you did not see her you could still, given the description she gave, have heard any slight reactions she might have involuntarily made, such as a slight cougn, or movement in her chair. So, even though I could not see her facial expressions or body language. Even though she uttered NOT a sound, she could still communicate to the listener what they were hearing. I wasn't aware that mental telepathy was a human characteristic. Given that you could not observe her reactions, that's still single blind because she could observe your reactions. If you could see or hear her reactions to the track number then it would just be a busted test. Obviously, she telepathically transmitted this data. IMHO, you are grasping straws here. |
#212
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Ed Seedhouse" wrote in message
... On May 21, 7:56 am, Sonnova wrote: The test administrator's knowledge of where the switch was. That's all it would take. There is no "switch". Certainly there was, given that your description of the process is accurate. The CD player was set to "shuffle" among the cuts. And there it is. All she knew was what cut NUMBER was playing, There's the knowledge of the switch. If a track number was repeated then she also had knowledge of your previous reaction to that particular track, which could have influenced you when the track repeated since she would have expectations based on your previous reaction which could have affected her. She might easily have given a facial or body language clue to you that might easily have influenced you. Even if you did not see her you could still, given the description she gave, have heard any slight reactions she might have involuntarily made, such as a slight cougn, or movement in her chair. Given that you could not observe her reactions, that's still single blind because she could observe your reactions. If you could see or hear her reactions to the track number then it would just be a busted test. This is the farce of the "objectivists"....this kind of grasping at straws, substituting the "possible" for the "probably". We've been over this here many times befor....that fact that it is remotely possible does not make it probable and invalidate the test. No more so than the fact that ANYTIME somebody wants to get "null" results with such a test they can just randomly guess. You say it won't happen. I say it is possible. In your world that invalidates the test ain a. This is at least, if not moreso, possibility than the a disinterested woman standing behind him and recording tracks and comments would be. |
#213
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 21, 11:58*am, "Harry Lavo" wrote:
"Steven Sullivan" wrote in message ... It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. If you are a scientist. *Which you are not. I find it interesting that no Sonnova, no Harry Lavo give any information about mp3 files that they use in their arguments. I happened to listen 32kbps tracks from Internet radio (classical music) and can assure you that you don't need any training or high quality system to hear artifacts of compression. At the same time I think that 320kbps mp3 encoded by iTunes is very hard if possible at all to distinguish from original CD. So, it is possible that Sonnova and Harry intentionally use low bit coded mp3 to make their point and people on another side argue merits of high bit competently encoded mp3's. Both sides are right in their domain and argument does not make sense at all. At the same time if Sonnova and Harry will tell us what they mean by mp3 then argument about encoding artifacts will become real. We can substitute DBT by the following experiment. Take a track from classical CD and make few files out of it: original, lossless, 320kbps and 256kbps. There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. Thc vlad |
#214
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 21, 3:01*pm, "Harry Lavo" wrote:
This is the farce of the "objectivists"....this kind of grasping at straws, substituting the "possible" for the "probably". *We've been over this here many times befor....that fact that it is remotely possible does not make it probable and invalidate the test. *No more so than the fact that ANYTIME somebody wants to get "null" results with such a test they can just randomly guess. *You say it won't happen. *I say it is possible. *In your world that invalidates the test ain a. *This is at least, if not moreso, possibility than the a disinterested woman standing behind him and recording tracks and comments would be. Actually the only claim I have ever made about the test under discussion was that so far as I am concerned it was not double blind, as the tester claimed. Your response of attributing thoughts, attitudes, beliefs and ideas to me that I do not have and have never expressed here or elsewhere is, alas, not surprising I am sorry to say. I am sad, however, that the moderator apparently thought your response was not a personal attack. I must say that it certainly strikes me as that. |
#215
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
vlad wrote:
We can substitute DBT by the following experiment. Take a track from classical CD and make few files out of it: original, lossless, 320kbps and 256kbps. There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. You also have to carefully time-align them and make sure they are equal length. Finally, even with th greatest of care, loading into Audacity and looking at the spectrum is almost always a dead giveaway. I've stupefied two people who tried what you described by telling them perfectly in which order (of bitrate) their re-.wavified MP3s were coded, just using the length and Audacity-done spectrum. It was pathetically easy. Of course, I could not tell what the actual MP3 bitrates were, just which was higher and which lower. Doug McDonald |
#216
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 22, 8:43*am, Ed Seedhouse wrote:
On May 21, 3:01*pm, "Harry Lavo" wrote: This is the farce of the "objectivists"....this kind of grasping at straws, substituting the "possible" for the "probably". *We've been over this here many times befor....that fact that it is remotely possible does not make it probable and invalidate the test. *No more so than the fact that ANYTIME somebody wants to get "null" results with such a test they can just randomly guess. *You say it won't happen. *I say it is possible. *In your world that invalidates the test ain a. *This is at least, if not moreso, possibility than the a disinterested woman standing behind him and recording tracks and comments would be. Actually the only claim I have ever made about the test under discussion was that so far as I am concerned it was not double blind, as the tester claimed. *Your response of attributing thoughts, attitudes, beliefs and ideas to me that I do not have and have never expressed here or elsewhere is, alas, not surprising I am sorry to say. actually you went much further than *just* claiming that the test was not double blind. You used that point as a means of trying to discredit the results. Here is the post an what you actually said On May 17, 11:39 am, Sonnova wrote: "If I took part in that test and had the same results described it would not convince me that a difference actually existed. Too many procedural problems. " So it would appear that Harry's comments do strike at the heart of this particular disagreement. It does seem that some how that which is Possible yet very unlikely has been used as probable cause for dismisal of results. |
#217
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 22, 8:45*am, "mcdonaldREMOVE TO ACTUALLY REACH
wrote: vlad wrote: We can substitute DBT by the following experiment. Take a track from classical CD and make few files out of it: original, lossless, 320kbps and 256kbps. There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. You also have to carefully time-align them and make sure they are equal length. Finally, even with th greatest of care, loading into Audacity and looking at the spectrum is almost always a dead giveaway. I've stupefied two people who tried what you described by telling them perfectly in which order (of bitrate) their re-.wavified MP3s were coded, just using the length and Audacity-done spectrum. It was pathetically easy. Of course, I could not tell what the actual MP3 bitrates were, just which was higher and which lower. Doug McDonald Totally agree with you. With usage of special tools/programs you can distinguish files given. But in DBT we all assume that the listener relies on his ears and only his ears. If somebody would go into a hassle of preparing these files and Sonnova would agree to this kind of test then there would be a gentlemanly agreement that no tools of any kind would be used except his ears. He would listen tracks on the system of his choice and then report to us his findings. BTW, I would expect that Sonnova and/or Harry would provide data about mp3 files that they mentioned in this forum. If it was 16kbps compressed mp3's then there is no argument indeed :-) Thx vlad |
#218
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 22, 10:19*am, wrote:
On May 22, 8:43*am, Ed Seedhouse wrote: Actually the only claim I have ever made about the test under discussion was that so far as I am concerned it was not double blind, as the tester claimed. * actually you went much further than *just* claiming that the test was not double blind. You used that point as a means of trying to discredit the results. That it was not in fact double blind and you claimed is what discredits (in my opinion) the results in the sense that the test referred to gives no evidence to the dispute one way or the other. Here is the post an what you actually said On May 17, 11:39 am, Sonnova wrote: "If I took part in that test and had the same results described it would not convince me that a difference actually existed. *Too many procedural problems. " Which is perfectly true and remains so. I have no idea why that should strike you as being a personal attack. So it would appear that Harry's comments do strike at the heart of this particular disagreement. To the contrary I would say. They seem to me to miss it entirely. Nowhere did I do any of the things that Harry attributed to "objectivists" in the guise, I believe, of attacking me. That you think it did gives evidence that we are talking past one another and should probably just shut up. I have made my point now and will not go further. |
#219
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Ed Seedhouse" wrote in message
... On May 21, 3:01 pm, "Harry Lavo" wrote: This is the farce of the "objectivists"....this kind of grasping at straws, substituting the "possible" for the "probably". We've been over this here many times befor....that fact that it is remotely possible does not make it probable and invalidate the test. No more so than the fact that ANYTIME somebody wants to get "null" results with such a test they can just randomly guess. You say it won't happen. I say it is possible. In your world that invalidates the test ain a. This is at least, if not moreso, possibility than the a disinterested woman standing behind him and recording tracks and comments would be. Actually the only claim I have ever made about the test under discussion was that so far as I am concerned it was not double blind, as the tester claimed. Your response of attributing thoughts, attitudes, beliefs and ideas to me that I do not have and have never expressed here or elsewhere is, alas, not surprising I am sorry to say. I am sad, however, that the moderator apparently thought your response was not a personal attack. I must say that it certainly strikes me as that. You also said, I believe, that it failed to be a proper test due to many possible objections, but indicated probable transmission of "cues" as the reason for the result. These were not disinterested observations, but direct challenges to Sonnova and his claimed results. If I am ascribing a second person's opinions to you, I apologize. But the preceding is what I recall from thie beginning of the thread. |
#220
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 22, 6:56*pm, "Harry Lavo" wrote:
"Ed Seedhouse" wrote in message ... On May 21, 3:01 pm, "Harry Lavo" wrote: This is the farce of the "objectivists"....this kind of grasping at straws, substituting the "possible" for the "probably". We've been over this here many times befor....that fact that it is remotely possible does not make it probable and invalidate the test. No more so than the fact that ANYTIME somebody wants to get "null" results with such a test they can just randomly guess. You say it won't happen. I say it is possible. In your world that invalidates the test ain a. This is at least, if not moreso, possibility than the a disinterested woman standing behind him and recording tracks and comments would be. Actually the only claim I have ever made about the test under discussion was that so far as I am concerned it was not double blind, as the tester claimed. *Your response of attributing thoughts, attitudes, beliefs and ideas to me that I do not have and have never expressed here or elsewhere is, alas, not surprising I am sorry to say. I am sad, however, that the moderator apparently thought your response was not a personal attack. *I must say that it certainly strikes me as that. You also said, I believe, that it failed to be a proper test due to many possible objections, but indicated probable transmission of "cues" as the reason for the result. *These were not disinterested observations, but direct challenges to Sonnova and his claimed results. *If I am ascribing a second person's opinions to you, I apologize. *But the preceding is what I recall from thie beginning of the thread. Then perhaps you will read more closely in future. For one thing, the word "probable" is not the same as "possible", which is the one I used. Nor did I dispute the results recorded, merely their status as a "double blind" test. I took Sonnova's description at his word and still do. Since it was not, in my opinion, a double blind test it has, to me, no evidenciary value, and does not change my opinion. Whether or not it changes any one else's is up to them. I think it highly likely that my opinion would be shared by a reviewer in a peer reviewed journal, but I cannot prove that of course. None of this was to denigrate anyone - to believe that someone is wrong is not to believe that they are bad. I am quite sure that both you and Sonnova hold perfectly sincere opinions on this matter. I think yours is, in this case, mistaken and you think mine is mistaken. Surely adults can disagree without being disagreeable. |
#221
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"vlad" wrote in message
... On May 22, 8:45 am, "mcdonaldREMOVE TO ACTUALLY REACH wrote: vlad wrote: We can substitute DBT by the following experiment. Take a track from classical CD and make few files out of it: original, lossless, 320kbps and 256kbps. There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. You also have to carefully time-align them and make sure they are equal length. Finally, even with th greatest of care, loading into Audacity and looking at the spectrum is almost always a dead giveaway. I've stupefied two people who tried what you described by telling them perfectly in which order (of bitrate) their re-.wavified MP3s were coded, just using the length and Audacity-done spectrum. It was pathetically easy. Of course, I could not tell what the actual MP3 bitrates were, just which was higher and which lower. Doug McDonald Totally agree with you. With usage of special tools/programs you can distinguish files given. But in DBT we all assume that the listener relies on his ears and only his ears. If somebody would go into a hassle of preparing these files and Sonnova would agree to this kind of test then there would be a gentlemanly agreement that no tools of any kind would be used except his ears. He would listen tracks on the system of his choice and then report to us his findings. BTW, I would expect that Sonnova and/or Harry would provide data about mp3 files that they mentioned in this forum. If it was 16kbps compressed mp3's then there is no argument indeed :-) For the record let it be said here that I have reported no listening test nor mentioned any specific mp3 bitrates or codecs in this discussion. The only thing I have done is come to Sonnava's defense in terms of the insinuation that his test was woefully inadequate. It was not; it may not have been perfect, but as he described it is was more "perfect" (i.e. double blind) than most such tests done in the home. |
#222
Posted to rec.audio.high-end
|
|||
|
|||
A blatantly positive MP3 test: was Is flat frequency response
I tried a test designed to allow me to hear MP3 artifacts
clearly. It succeeded. But it required a truly ghastly set of kludges. What I did was take some music (a Mozart piano concerto excerpt) and high-pass filter it at 4 kHz. This was done in Audacity with several passes the the equalizer, to lower the level by 90 dB (using 32 bit float numbers). The peaks were normalized to 30% full scale. Then I created a 440 Hz tone at also 30% full scale. I added the tracks together and saved as a wave file. That was converted to fixed bitrate 128bps using LAME. I then reloaded the MP3 and was barely able to hear a difference when the high frequency peaks occurred (mostly, of course, during the soft parts, there was no content above 4 kHz. Low pass filtering this at 3 kHz made things more obvious since there now were no high frequencies there at all. Finally, I subtracted off the sine wave to get it down some 30 dB, and the artifacts were perfectly blatant, sounding exactly like what I had expected. This was about the toughest possible test, and at 128 kbps the errors were obvious. But is is clear why it is so hard for me to tell on real music. Doug McDonald |
#223
Posted to rec.audio.high-end
|
|||
|
|||
A blatantly positive MP3 test: was Is flat frequency response
"mcdonaldREMOVE TO ACTUALLY REACH wrote:
I tried a test designed to allow me to hear MP3 artifacts clearly. It succeeded. But it required a truly ghastly set of kludges. What I did was take some music (a Mozart piano concerto excerpt) and high-pass filter it at 4 kHz. This was done in Audacity with several passes the the equalizer, to lower the level by 90 dB (using 32 bit float numbers). The peaks were normalized to 30% full scale. Then I created a 440 Hz tone at also 30% full scale. I added the tracks together and saved as a wave file. That was converted to fixed bitrate 128bps using LAME. I then reloaded the MP3 and was barely able to hear a difference when the high frequency peaks occurred (mostly, of course, during the soft parts, there was no content above 4 kHz. Low pass filtering this at 3 kHz made things more obvious since there now were no high frequencies there at all. Finally, I subtracted off the sine wave to get it down some 30 dB, and the artifacts were perfectly blatant, sounding exactly like what I had expected. Then I redid using LAME at VBR 3, which averages about 190 kbps. The sine wave was essentially perfect. No difference between the original and the reconstruction was audible TO ME no matter how I filtered. Doug McDonald |
#224
Posted to rec.audio.high-end
|
|||
|
|||
A blatantly positive MP3 test: was Is flat frequency response
"mcdonaldREMOVE wrote:
I tried a test designed to allow me to hear MP3 artifacts clearly. It succeeded. But it required a truly ghastly set of kludges. why not just encode at a low bitrate? Or try the various sets of 'killer' samples for mp3 tuning, eg: http://lame.sourceforge.net/quality.php http://ff123.net/samples.html -- -S We have it in our power to begin the world over again - Thomas Paine |
#225
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? Well, I listened myself for as long as I could stand it on my "very fine" equipment. Not good enough (your 'education',not your audio gear) antSo I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of anecdotal evidence around. Not good enough. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. Nonsense, Harry. What 'digital noise from a PC' are you referring to and why do you insist it's both common and audible? I'm not arguing that headphones can't demonstrate certain distortions...just that they have never been tested to show that they can demonstrate ALL distortions. They can certainly demonsrate low-level differences that would not usually be apparent in 'open air' listening...one typical reasonn being the higher S/N . It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. If you are a scientist. Which you are not. Actually, I am, with a PhD and publications and everything, Harry. A biologist, to be precise. So I'm going to have to ask you to retract that claim. -- -S We have it in our power to begin the world over again - Thomas Paine |
#226
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
"Steven Sullivan" wrote in message ... Arny Krueger wrote: "Harry Lavo" wrote in message It's also not automatic, Steven. Folks worked for years to get compression schemes that were even acceptable to a large majority of trained listeners. "large majority of trained listeners" - an undefined group if there ever was one. I have no idea who he is even talking about. There aren't that many trained listeners for mp3 artifacts. JJ and others have indicated that listeners are first screened for basic accuity, and then must be specifically trained to recognize and differentiate the artifact under question before a valid test can be done. That's what a trained listener is/was. In the early days, this led to more tests getting confirmed as different than did tests that indicated "no difference". True, but we are *well* past the early days. In fact, most LAME codec tweaking has been done using computer speakers and headphones. So perhaps they missed the kind that Sonnova picked up? How to you test "air" on a headphone? When phase is largely taken out of the picture? It wasn't just headohones,'largely' does not mean 'all', and "perhaps" is not strong enough. And why *would* phase artifacts be any more audible than others, in high-bitrate mps --which Sonnova says he can ALWAYS identify. If you have good ABX test results,feel free to report them. -- -S We have it in our power to begin the world over again - Thomas Paine |
#227
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
vlad wrote:
On May 21, 11:58?am, "Harry Lavo" wrote: "Steven Sullivan" wrote in message ... It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. If you are a scientist. ?Which you are not. I find it interesting that no Sonnova, no Harry Lavo give any information about mp3 files that they use in their arguments. I happened to listen 32kbps tracks from Internet radio (classical music) and can assure you that you don't need any training or high quality system to hear artifacts of compression. At the same time I think that 320kbps mp3 encoded by iTunes is very hard if possible at all to distinguish from original CD. That's rather odd. What encoder was uses for the 320 kbps downloads you can hear the compression artifacts in? We can substitute DBT by the following experiment. Take a track from classical CD and make few files out of it: original, lossless, 320kbps and 256kbps. Or, use variable bitrate with a ~190 average. Just be sure to use the current LAME (or AAC , 256) encoder, as recommended by the people who, uinlike Harry or Sonnova, actually work with and test mp3s regularly.. There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. Which is the test I did with a handful of skeptics, and none of those who reported back performed better than chance. The only thing beaware of is the possibility of the mp3--wav introducing clipping, if the peaks were already near 0 dBFS. So check the wavs for that. -- -S We have it in our power to begin the world over again - Thomas Paine |
#228
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Steven Sullivan wrote:
There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. Which is the test I did with a handful of skeptics, and none of those who reported back performed better than chance. You have to be careful of cheating. I'm very good at that with these "send me the files" tests. It's a pain to do, much more time consuming than listening, but it usually works and tells which of the encode-decode-send-the- decoded-one files used more, and which fewer, bits. I just load into Audaciy, subtract the original, and look at the spectrum. Doug McDonald |
#229
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 21, 2:58*pm, "Harry Lavo" wrote:
snip Well, I listened myself for as long as I could stand it on my "very fine" equipment. *So I tend to support those who hear a difference on their "very fine" equipment, among other things. *But there is certainly lots of antecdotal evidence around. *An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. * Please explain what you mean about "bathed in digital noise from a PC". How does such noise sound? Many commercial recordings are made using PCs and Macs. Do they suffer from being bathed in digital noise? If you are a scientist. *Which you are not. IIRC, Dr. Sullivan has a Ph.D. in biochemistry. Most people would disagree with you here Harry. |
#230
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Steven Sullivan" wrote in message
... Harry Lavo wrote: Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? Well, I listened myself for as long as I could stand it on my "very fine" equipment. Not good enough (your 'education',not your audio gear) antSo I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of anecdotal evidence around. Not good enough. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. Nonsense, Harry. What 'digital noise from a PC' are you referring to and why do you insist it's both common and audible? You ever hear the effect of a PC on a good audio system when it is operating nearby? Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. I'm not arguing that headphones can't demonstrate certain distortions...just that they have never been tested to show that they can demonstrate ALL distortions. They can certainly demonsrate low-level differences that would not usually be apparent in 'open air' listening...one typical reasonn being the higher S/N . Yep, which is why I presume the developers of codecs found them useful. But as I said, that still doesn't mean that headphones revewal all the subtle imaging effects that we are used to as audiophiles. And there is no reason to not believe that codecs can't effect that as well. It is one thing to be skeptical...it is another to automatically say somebody is an "outlier" and not to be believed because they report results that are not the norm. It is utterly proper to call a rare result -- both predicted to be rare, and rare in practice -- an 'outlier'. If you are a scientist. Which you are not. Actually, I am, with a PhD and publications and everything, Harry. A biologist, to be precise. So I'm going to have to ask you to retract that claim. I certainly will retract that....for all the years that you and I have posted here, I somehow missed that. I apologize. |
#231
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On May 25, 8:15*pm, "Harry Lavo" wrote:
snip You ever hear the effect of a PC on a good audio system when it is operating nearby? *Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. *Even if at a very low, subliminal-like level, it still affects the audio signal. *Not exactly high-fidelity. One might suggest that such an audio system is actually quite deficient if a nearby PC degrades the performance. Take a look at the Lynx 22 converter with a dynamic range of 115 dB. This is a PCI card so it resides in this supposedly horrible environment but being well designed it works quite well. Even some low-cost sound cards can have a dynamic range greater than 80 dB which is well below the background noise of typical recording environments. As noted elsewhere, many commercial recordings are made using PCs and Macs without any apparent deficiencies. |
#232
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Harry Lavo" wrote in message
You ever hear the effect of a PC on a good audio system when it is operating nearby? Of course, but how much depends on the PC. Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typical cheap analog circuitry built into most pcs. Sometimes yes, often no. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. Here we have someone who clearly is propounding the myths that: (1) All PCs sound the same. (bad) (2) There's no way to manage any of the problems that some PCs have. I'm not arguing that headphones can't demonstrate certain distortions...just that they have never been tested to show that they can demonstrate ALL distortions. They can certainly demonstrate low-level differences that would not usually be apparent in 'open air' listening...one typical reason being the higher S/N . Yep, which is why I presume the developers of codecs found them useful. But as I said, that still doesn't mean that headphones reveal all the subtle imaging effects that we are used to as audiophiles. More to the point, loudspeakers can't possibly reveal all of the subtle imaging effects that are audible via headphones, because of all the extraneous garbage that loudspeakers and rooms bring into the listening situation. And there is no reason to not believe that codecs can't effect that as well. Indeed, that's why people do careful listening tests to see exactly what given codecs are doing. At this point it is arguable that codecs are among to most carefully-listened-to audio components around. Certainly, codecs have been more carefully listened to than say power amplifiers, since almost nobody does good DBTs of amplifiers. |
#233
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
You ever hear the effect of a PC on a good audio system when it is operating nearby? Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. You REALLY need to get out more. The Lynx 2 is one of the quietest audio interfaces that there is. A LOT more quiet than most so-called stuff that's labeled "high-end" which is often pretty noisy in comparison. I know - I've tested a lot of it and this is provable, both objectively and empirically. Funny (in the strange sense) that you don't mention the acoustic noise from the fans, which IS a real problem, but solvable. I get around it by using a fanless power supply, an efficient low speed fan for the CPU and putting the machine in a nearby closet. |
#234
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
wrote in message
On May 21, 2:58 pm, "Harry Lavo" wrote: snip Well, I listened myself for as long as I could stand it on my "very fine" equipment. So I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of antecdotal evidence around. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. Please explain what you mean about "bathed in digital noise from a PC". How does such noise sound? Many commercial recordings are made using PCs and Macs. Do they suffer from being bathed in digital noise? The myth is that PCs are somehow uniquely bathed in digital noise. Many kinds of modern audio components are bathed in digital noise, such as DACs, optical disc players, and surround decoders and receivers, but nobody seems to be very worried about them. There is a science and art called "mixed signal design". Executed well, it can make modern digital/analog equipment effectively noise free. |
#235
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
wrote in message ...
Harry Lavo wrote: You ever hear the effect of a PC on a good audio system when it is operating nearby? Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. You REALLY need to get out more. The Lynx 2 is one of the quietest audio interfaces that there is. A LOT more quiet than most so-called stuff that's labeled "high-end" which is often pretty noisy in comparison. I know - I've tested a lot of it and this is provable, both objectively and empirically. Funny (in the strange sense) that you don't mention the acoustic noise from the fans, which IS a real problem, but solvable. I get around it by using a fanless power supply, an efficient low speed fan for the CPU and putting the machine in a nearby closet. And how many audiophiles with very good main systems have a computer with a Lynx sound card? If they record live, maybe. Otherwise, probably not. I'm talking about ordinary people with ordinary computers doing a computerized ABX or ABC/hr test as per the POV expressed here that that is somehow superior to listening on one's main high-end system. |
#236
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
And how many audiophiles with very good main systems have a computer with a Lynx sound card? If they record live, maybe. Otherwise, probably not. I'm talking about ordinary people with ordinary computers doing a computerized ABX or ABC/hr test as per the POV expressed here that that is somehow superior to listening on one's main high-end system. You mentioned "high fidelity" in your post, so I was responding to that. |
#237
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"Harry Lavo" wrote in message
And how many audiophiles with very good main systems have a computer with a Lynx sound card? Easily thousands. If they record live, maybe. I owned a LynxOne and a Lynxtwo for a number of years before I did any serious recording. And before that, a CardDeluxe. Otherwise, probably not. I'm talking about ordinary people with ordinary computers doing a computerized ABX or ABC/hr test as per the POV expressed here that that is somehow superior to listening on one's main high-end system. The point is Harry that all it takes is about $100 for an Audiophile 2496, and you've got a PC playback system that will equal or beat any CD player. Due to the continuing drop in the price of good converter chips, for than $200 you can get one several interfaces from eMu that will match the performance of a LynxTwo. |
#238
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
On Mon, 25 May 2009 17:15:05 -0700, Harry Lavo wrote
(in article ): "Steven Sullivan" wrote in message ... Harry Lavo wrote: Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? Well, I listened myself for as long as I could stand it on my "very fine" equipment. Not good enough (your 'education',not your audio gear) antSo I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of anecdotal evidence around. Not good enough. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. Nonsense, Harry. What 'digital noise from a PC' are you referring to and why do you insist it's both common and audible? You ever hear the effect of a PC on a good audio system when it is operating nearby? Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. My understanding is that on well designed sound-cards it's not a problem. Sensitive components are well shielded. I have an Apple TV box which is a complete computer with sound components and it is connected to my video system and even with fairly decent speakers connected and with the volume pretty loud, I hear NOTHING but the program. Even with nothing playing there is no discernable noise. snip |
#239
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
"mcdonaldREMOVE wrote:
Steven Sullivan wrote: There is an issue, of course, of file size. But if we will restore each of compressed files to straight CD encoding then they will be pretty much of the same size. Make few copies of each, end send them to Sonnova. I wonder if he will be able to pinpoint compressed files reliably above simple guessing. Which is the test I did with a handful of skeptics, and none of those who reported back performed better than chance. You have to be careful of cheating. Indeed, and they *could* have cheated by analysing the file spectra...but apparently none did. -- -S We have it in our power to begin the world over again - Thomas Paine |
#240
Posted to rec.audio.high-end
|
|||
|
|||
Is flat frequency response desirable?
Harry Lavo wrote:
"Steven Sullivan" wrote in message ... Harry Lavo wrote: Who says 'fine equipment' makes mp3 artifacst more audible? Really, Harry, where have you gotten your information about mp3s from? Well, I listened myself for as long as I could stand it on my "very fine" equipment. Not good enough (your 'education',not your audio gear) antSo I tend to support those who hear a difference on their "very fine" equipment, among other things. But there is certainly lots of anecdotal evidence around. Not good enough. An in-room stereo of good quality reveals much more than headphones, especially ones bathed in digital noise from a PC, when it comes to imaging, image depth, the "realism" of the sound. Nonsense, Harry. What 'digital noise from a PC' are you referring to and why do you insist it's both common and audible? You ever hear the effect of a PC on a good audio system when it is operating nearby? No. In fact, I play much of my music library *through* a PC (a laptop) feeding digital out via USB to an AVR with USB input. Do you think that same low level noise does not invade the analog circuitry in the PC itself, feeding the audio/headphones out on the typlical cheap analog circuitry built into most pcs. Even if at a very low, subliminal-like level, it still affects the audio signal. Not exactly high-fidelity. I think you need to prove that what you claim to be a common audible problem, is. be apparent in 'open air' listening...one typical reasonn being the higher S/N . Yep, which is why I presume the developers of codecs found them useful. But as I said, that still doesn't mean that headphones revewal all the subtle imaging effects that we are used to as audiophiles. And there is no reason to not believe that codecs can't effect that as well. Of course codecs *can* affect lots of things. The issue is whether they audibly DO, and under what conditiosn. You haven't even begun to show that you've investigated the issue enough to make a claim one way or the other. Actually, I am, with a PhD and publications and everything, Harry. A biologist, to be precise. So I'm going to have to ask you to retract that claim. I certainly will retract that....for all the years that you and I have posted here, I somehow missed that. I apologize. Accepted. -- -S We have it in our power to begin the world over again - Thomas Paine |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
Why are mic made with non flat frequency response? | Pro Audio | |||
Frequency Response of XM | High End Audio | |||
Frequency response | Pro Audio | |||
Frequency response | Pro Audio | |||
Mic Frequency Response | Pro Audio |