syn08 said:
audio != medicine
The results in medicine are empirically evident. Not in audio.
mathematics and statistics for a DBT are meaningful ONLY with good data.
THE question is if any DBTs in audio have produced any meaningful data!
Merely doing a DBT does not guarantee any results that have any gravitas.
Can't speak for Curl, but as I stated, where is the "science" in tests that can not be properly replicated because sufficient data and controls are not in place or reported?
The issue is not IF DBT testing is a valid or useful technique. Rather the question is if DBT is to be used for audio how can it be employed to produce results that are both meaninful and generalizable beyond the specific tests.
And by generalizable beyond the specific tests I mean a) reproducable results (that other researchers verify) and b) results that really do apply to what we can term SOTA high resolution playback chains - not just "good high-end" or "blameless" playback chains.
It's not quite as simple as it looks on the surface.
Please keep in mind that some very well trained "engineers" and the like argued that anything with a distortion level of <0.001% was going to be both sonically clean beyond audible thresholds and indistinguishable from any other thing of similar specs.
Ask urself if all CD playback systems sound the same to you or not? They all (typically) fall below that "0.001% distortion" level, right?
_-_-bear
Care to give some examples from JAES or JAS? The ones I've seen were very well controlled and show terrific sensitivity to exeedingly minor effects of level, frequency response, polarity, and in some cases phase. I've participated in a few, most recently a test of data compression (correctly distinguishing four variables).
The tests that many seem to like are those that have no control whatever. Either full knowledge of what's being listened to at any given time or at best no double blind and level matching. I'm sure the Blowtorch is wonderful (and unlike most in this thread, I've actually listened to one), but I'm unconvinced that sonically it's any more wonderful than any other preamp John has designed. Maybe it is, I don't know. In my own experience, line amps are pretty easy to make totally transparent- the difficulties are with power amps and phono stages. (I'm excluding transducers here, which are wildly variable and very easy to distinguish in controlled DBTs)
Originally posted by bear
audio != medicine
The results in medicine are empirically evident. Not in audio.
To start with, you lost me completely here.
mathematics and statistics for a DBT are meaningful ONLY with good data.
Mathematics offers methods to verify the validity of data. A good book on statistics will explain how.
THE question is if any DBTs in audio have produced any meaningful data!
DBT is NOT producing data, DBT offers a consistent and mathematically/statistically correct way to verify a hypothesis. Again, a book on statistics would help with the details.
Merely doing a DBT does not guarantee any results that have any gravitas.
The results of DBT are guaranteed with a certain probability. As I've said, if you are expecting certitudes from a statistical test, then you are fooling yourself.
Can't speak for Curl, but as I stated, where is the "science" in tests that can not be properly replicated because sufficient data and controls are not in place or reported?
Have you tried to replicate a DBT test and got contradictory results? Do I hear a No? And otherwise who decides if a test "can not be properly replicated"? Certainly not you or JC, but people with a solid background in statistics.
The issue is not IF DBT testing is a valid or useful technique. Rather the question is if DBT is to be used for audio how can it be employed to produce results that are both meaninful and generalizable beyond the specific tests.
And by generalizable beyond the specific tests I mean a) reproducable results (that other researchers verify) and b) results that really do apply to what we can term SOTA high resolution playback chains - not just "good high-end" or "blameless" playback chains.
Mind you again, this is mathematics and terminology has to be defined. Please define what you mean by "meaningful", "generalizable" and "beyond the specific tests" means, in the context of statistical analysis. To me, every correctly designed and executed DBT test results are meaningful and generalizable. Of course, I know how to read them.
It's not quite as simple as it looks on the surface.
I agree on this one
Please keep in mind that some very well trained "engineers" and the like argued that anything with a distortion level of <0.001% was going to be both sonically clean beyond audible thresholds and indistinguishable from any other thing of similar specs.
I don't see any relation with DBT here. But a DBT to test the validity (with, say, 99.996% probability) of the hypothesis "distortions under 0.001% are too small to be audible" can be designed, executed and reproduced by anybody. Give it a shot and let us know the results. If you are in the 0.004% that can hear a difference, this does not invalidate the test, but only shows your belonging to the Gifted People subset.
Ask urself if all CD playback systems sound the same to you or not? They all (typically) fall below that "0.001% distortion" level, right?
No, I'm saying all CD players sound the same for, say, 99.99% people. Therefore, the probability to hear a sound improvement after spending $15,000 on a CD transport, a supa dupa DAC and a Blowtorch is virtually nil. Therefore, it would make sense to spend that kind of money only if one can confirm his hearing abilities. Which again tells nothing about what I, or you, or Scott, or JC can hear, but tells that the majority of humans won't hear a damn thing.
Which boils down to one of the issues in the whole stinking high end audio business: they are selling equipment for obscene prices without knowing anything about the customer and if he's going to hear what he is promised. In a sense, they sell illusions not equipment.
Don't take me wrong: I'm myself occasionally bitten by the high end bug: I just purchased a SME Series III tonearm, to use with a high compliance cartridge, although I know damned well it won't make any difference to my ears, comparing to my current Audioquest PT-9 tonearm. But I like to experiment, and I also like the look and feel of the SME stuff. In fact, I know what I am doing: I'm gambling and having fun
Though, you won't find me writing for any media "my new SME Series III tonearm sounded energetic, lively, and eager to communicate".
audio != medicine
The results in medicine are empirically evident. Not in audio.
To start with, you lost me completely here.
mathematics and statistics for a DBT are meaningful ONLY with good data.
Mathematics offers methods to verify the validity of data. A good book on statistics will explain how.
THE question is if any DBTs in audio have produced any meaningful data!
DBT is NOT producing data, DBT offers a consistent and mathematically/statistically correct way to verify a hypothesis. Again, a book on statistics would help with the details.
Merely doing a DBT does not guarantee any results that have any gravitas.
The results of DBT are guaranteed with a certain probability. As I've said, if you are expecting certitudes from a statistical test, then you are fooling yourself.
Can't speak for Curl, but as I stated, where is the "science" in tests that can not be properly replicated because sufficient data and controls are not in place or reported?
Have you tried to replicate a DBT test and got contradictory results? Do I hear a No? And otherwise who decides if a test "can not be properly replicated"? Certainly not you or JC, but people with a solid background in statistics.
The issue is not IF DBT testing is a valid or useful technique. Rather the question is if DBT is to be used for audio how can it be employed to produce results that are both meaninful and generalizable beyond the specific tests.
And by generalizable beyond the specific tests I mean a) reproducable results (that other researchers verify) and b) results that really do apply to what we can term SOTA high resolution playback chains - not just "good high-end" or "blameless" playback chains.
Mind you again, this is mathematics and terminology has to be defined. Please define what you mean by "meaningful", "generalizable" and "beyond the specific tests" means, in the context of statistical analysis. To me, every correctly designed and executed DBT test results are meaningful and generalizable. Of course, I know how to read them.
It's not quite as simple as it looks on the surface.
I agree on this one
Please keep in mind that some very well trained "engineers" and the like argued that anything with a distortion level of <0.001% was going to be both sonically clean beyond audible thresholds and indistinguishable from any other thing of similar specs.
I don't see any relation with DBT here. But a DBT to test the validity (with, say, 99.996% probability) of the hypothesis "distortions under 0.001% are too small to be audible" can be designed, executed and reproduced by anybody. Give it a shot and let us know the results. If you are in the 0.004% that can hear a difference, this does not invalidate the test, but only shows your belonging to the Gifted People subset.
Ask urself if all CD playback systems sound the same to you or not? They all (typically) fall below that "0.001% distortion" level, right?
No, I'm saying all CD players sound the same for, say, 99.99% people. Therefore, the probability to hear a sound improvement after spending $15,000 on a CD transport, a supa dupa DAC and a Blowtorch is virtually nil. Therefore, it would make sense to spend that kind of money only if one can confirm his hearing abilities. Which again tells nothing about what I, or you, or Scott, or JC can hear, but tells that the majority of humans won't hear a damn thing.
Which boils down to one of the issues in the whole stinking high end audio business: they are selling equipment for obscene prices without knowing anything about the customer and if he's going to hear what he is promised. In a sense, they sell illusions not equipment.
Don't take me wrong: I'm myself occasionally bitten by the high end bug: I just purchased a SME Series III tonearm, to use with a high compliance cartridge, although I know damned well it won't make any difference to my ears, comparing to my current Audioquest PT-9 tonearm. But I like to experiment, and I also like the look and feel of the SME stuff. In fact, I know what I am doing: I'm gambling and having fun
Though, you won't find me writing for any media "my new SME Series III tonearm sounded energetic, lively, and eager to communicate".syn08 said:
I'm personally not sure that you could get anything in the audio world to be generally accepted, but how about this:
One of the nagging problems about double blind testing for sound reproduction questions is that the test instrument is a human being. In most (not all) medical testing, the human is part of the system under test. Some other detached indicator is used to measure the results. For example, a blood analysis. The researchers don't base their test results entirely on a patient's perception of whether his triglyceride level has fallen or not. In fact, if I understand right, people's emotional reaction to any variable in the test is so significant that placebo alternatives are included to sort this out.
So, what is needed is a way to actually measure the brain response of a listener to different sound reproduction systems. With proper testing techniques, the variables in peoples' perception mechanisms and most everything else could be isolated from the actual audio system changes.
Sounds nice, but not possible, right?
Well... This kind of thing is actually being done right now in laboratories for other types of responses. This work is carried out done by guys in white coats who have very advanced degrees and submit their work to peer reviewed journals and all that jazz. Perhaps somebody who is really motivated could get one of these researchers to try to resolve the burning question of whether and how audio blind testing really works.
Here is an example of one these researchers: http://www.ccnl.emory.edu/greg/
syn08 said:Originally posted by bear
audio != medicine
The results in medicine are empirically evident. Not in audio.
To start with, you lost me completely here.
I noticed, old boy!
mathematics and statistics for a DBT are meaningful ONLY with good data.
Mathematics offers methods to verify the validity of data. A good book on statistics will explain how.
you're mistaking the mathematical gyrations for the actual test and the recorded perceptions; that result being useful or valid - and reproducable
THE question is if any DBTs in audio have produced any meaningful data!
DBT is NOT producing data, DBT offers a consistent and mathematically/statistically correct way to verify a hypothesis. Again, a book on statistics would help with the details.
sure but only if what you are testing in the first place isn't utter nonsense
Merely doing a DBT does not guarantee any results that have any gravitas.
The results of DBT are guaranteed with a certain probability. As I've said, if you are expecting certitudes from a statistical test, then you are fooling yourself.
look, ur talking about statistics - I'm not. I grant you the statistics being a valid methodology. if you test for distortions using a speaker (or other componenent) that has higher distortions, noise levels, or other masking components, then what use are the results, or the "valid results"?
Can't speak for Curl, but as I stated, where is the "science" in tests that can not be properly replicated because sufficient data and controls are not in place or reported?
Have you tried to replicate a DBT test and got contradictory results? Do I hear a No? And otherwise who decides if a test "can not be properly replicated"? Certainly not you or JC, but people with a solid background in statistics.
eh? how can anyone replicate an audio DBT test accurately? Is there enough basic data regarding the setups provided? Has the performance of the gear been verified and published? how about the room? how about the switching mechanism used - does it have artifacts? what about the ground loop issues that jneutron discussed - where are the controls?? etc...
The issue is not IF DBT testing is a valid or useful technique. Rather the question is if DBT is to be used for audio how can it be employed to produce results that are both meaninful and generalizable beyond the specific tests.
And by generalizable beyond the specific tests I mean a) reproducable results (that other researchers verify) and b) results that really do apply to what we can term SOTA high resolution playback chains - not just "good high-end" or "blameless" playback chains.
Mind you again, this is mathematics and terminology has to be defined. Please define what you mean by "meaningful", "generalizable" and "beyond the specific tests" means, in the context of statistical analysis. To me, every correctly designed and executed DBT test results are meaningful and generalizable. Of course, I know how to read them.
so you can say that "all xyz are foo" as the result of a published DBT on audio that used valid statistical methods? no questions?
It's not quite as simple as it looks on the surface.
I agree on this one
Please keep in mind that some very well trained "engineers" and the like argued that anything with a distortion level of <0.001% was going to be both sonically clean beyond audible thresholds and indistinguishable from any other thing of similar specs.
I don't see any relation with DBT here. But a DBT to test the validity (with, say, 99.996% probability) of the hypothesis "distortions under 0.001% are too small to be audible" can be designed, executed and reproduced by anybody. Give it a shot and let us know the results. If you are in the 0.004% that can hear a difference, this does not invalidate the test, but only shows your belonging to the Gifted People subset.
DBT is a testing method. DBT for audio is a problematic method - many uncontrolled variables, thus often unclear results when plumbing the subtleties, which is where this entire thread, and much of the more sophisticated design efforts lie - like your complex and well thought out amp designs!
Ask urself if all CD playback systems sound the same to you or not? They all (typically) fall below that "0.001% distortion" level, right?
No, I'm saying all CD players sound the same for, say, 99.99% people. Therefore, the probability to hear a sound improvement after spending $15,000 on a CD transport, a supa dupa DAC and a Blowtorch is virtually nil. Therefore, it would make sense to spend that kind of money only if one can confirm his hearing abilities. Which again tells nothing about what I, or you, or Scott, or JC can hear, but tells that the majority of humans won't hear a damn thing.
Which boils down to one of the issues in the whole stinking high end audio business: they are selling equipment for obscene prices without knowing anything about the customer and if he's going to hear what he is promised. In a sense, they sell illusions not equipment.
sure... those who like with technological sculpture...
not the issue. we're dealing with the "truth" and with sophisticated designs and the limits of perception, not idiots with Bose Waveradiators (or whatever they are)
Don't take me wrong: I'm myself occasionally bitten by the high end bug: I just purchased a SME Series III tonearm, to use with a high compliance cartridge, although I know damned well it won't make any difference to my ears, comparing to my current Audioquest PT-9 tonearm. But I like to experiment, and I also like the look and feel of the SME stuff. In fact, I know what I am doing: I'm gambling and having fun
yeah? ya won't write it for the media, but I guarantee you will probably prefer one or the other, and maybe use one for one type of material and the other for other, exclusively... ha!
[/B] [/B]
Bear in mind also, that one of the more important aspects of high end ownership IS the tactile sensations a piece of equipment generates for the user, which has nothing to do with the measured performance of the equipment in question. Is doesn't matter how good a piece of equipment sounds to the user, if it looks like it was built from spare parts in someone's basement, as a product, it will fail.
Aesthetics play a crucial role in high end audio. Regardless of price, the tactile experience is at the top of the list of goodies the user expects to receive for their money. SME tonearms are nice to use, and the turntable one of the most tactile experiences audio has to offer. One of the many reasons Vinyl refuses to die. Also, one of the only reasons to visit Hot Topic.
Aesthetics play a crucial role in high end audio. Regardless of price, the tactile experience is at the top of the list of goodies the user expects to receive for their money. SME tonearms are nice to use, and the turntable one of the most tactile experiences audio has to offer. One of the many reasons Vinyl refuses to die. Also, one of the only reasons to visit Hot Topic.
Corvus corax said:
I absolutely agree with the above.
The only problem is not having anything to do with audio, sound, technology, electronics, etc... but with what is called "industrial design" and, of course, marketing. Now, if this would be a proffessional forum about how to build and sell commercial audio equipment (some are trying to use it in this way, though) this would be a very interesting topic. But this is DIY Audio...
Nothing wrong with making it look professional, though. That, and since pretty much any circuit we build is going to exhibit microphony to some degree, enclosure design is a very relevant topic indeed. This thread is about building superior sounding preamplifiers, is it not?
After the obvious subject of circuit design, the actual physical execution is the single most important factor in producing a high performance component, and so, is especially relevant to commercial as well as DIY components.
It is also what separates the serious DIY builder from the amateur. There are some really great examples of fine gear on these boards.
After the obvious subject of circuit design, the actual physical execution is the single most important factor in producing a high performance component, and so, is especially relevant to commercial as well as DIY components.
It is also what separates the serious DIY builder from the amateur. There are some really great examples of fine gear on these boards.
I predict that it will become 'un-cool' to claim to have Golden Ears in the future.
But before that happens I hope the guy who actually has the best ears in the world is discovered and he/she will agree to some DBTing.
He could be in this forum right now. Put yourself forward whilst it's still cool.
But before that happens I hope the guy who actually has the best ears in the world is discovered and he/she will agree to some DBTing.
He could be in this forum right now. Put yourself forward whilst it's still cool.
- Status
- Not open for further replies.