Zp 24/96

Agreed.

Personally, I cannot take anyone who claims the sonic benefits of hires are "obvious" seriously. That is simply not the case.

I'm still open minded about whether the differences are audible at all, or not but to claim they are "obvious" is simply incorrect. Even most hardened audiophiles will agree the differences are subtle and that you may require prolonged listening in a familiar environment (such as your own home) with familiar source material to detect the nuances.

I agree wholeheartedly. I think anyone saying it is obvious is simply blowing smoke.

At the end of the day, I am not even sure there is a perceptible difference.

I simply have tracks that *are* high resolution and I'd like to play then. Noticeable difference or not.

gtyer,

I agree with you. I don't think blind ABX is the be all/end all/last word in proving whether A is better than B. But what I hope when one states that A is better than B is that they have taken some effort to eliminate personal biases and are giving each of the choices a fair test.

Most of the time, I don't feel this happens with regard to comparing music/audio/electronics. Doesn't have too much to do with science. State the opinion that A is better than B, then tell us what you did to come to that conclusion. If it is stated, "I did such and such to eliminate variables, I did so and so to eliminate biases. I matched gain on both signals so that Fletcher-Munson would be happy...." then I can warm up to these opinions. But when it is stated that "even my deaf grandfather/indifferent significant other could hear the difference, why can't you?" I just have to take issue with that.

You intrigue me with your ideas of testing with brain monitoring. Would still seem to be fraught with error/inconclusiveness and subject to interpretation. If you could recommend some further reading, I am all ears, (unbiased ears I might add);)

You intrigue me with your ideas of testing with brain monitoring. Would still seem to be fraught with error/inconclusiveness and subject to interpretation. If you could recommend some further reading, I am all ears, (unbiased ears I might add);)

To exclude inconclusiveness - one would have to test all subjects at rest. I would assume baselines for all subjects would be different, so one would have to determine average baselines and individual subject baselines. (for all testing parameters)

Once the baselines are established - I would set up listening sessions.

The primary bias I'd be worried that could be introduced is "personal" bias. Basically, how the individual is feeling on that day. I'm sure music affects each person based on their mood. Since this test would be to determine the effects to the body while listening to music - it might be difficult to separate.

At the end of the day, ABX tests for changes to sensory input is invalid science. Like I've said - it's hard for the human memory to process change in senses. In some of these cases, the changes would be miniscule - AT BEST. So, in order for an even half-way valid ABX test, the subject has to be intimately familiar with the subject (in this case, it would be song). The subject would have to be intimately familiar with the environment (in this case the room/speaker set up). Then test. But that's near impossible.

Anyone that says hi-res music is ridiculously audibly different is, in my opinion, full of it. In that case an ABX test would have a high success rate. But it's a minor difference, if any at all.

Science aside, I'd simply like Sonos to be able to handle my music library. And since a lot of my library is vinyl recorded by a friend on his over-the-top setup at 24/96 ... I'd enjoy being able to feed that through. So, my desire is purely based on wanting my playlist available and nothing to do with quality.

I've been a party to lots of A/B showdowns and the results were mostly "my dog is better than your dog, period". No one wins and everyone goes home mad. Every once in a while there is wide agreement among panelists that one unit sounds better. I wish that I could put some science on that, but I can't. While we could agree which unit sounded better, we could not always agree on why it sounded better and that is troublesome.

I'm not so keen on the ABX box because many arrogantly accept that it is the only "neutral" box in existence. Why is that? If we can fabricate an ABX box that does not affect the sound quality, why can't we fabricate equipment that has this quality too? I my experience the ABX box obscures the differences between units. I've worked with units that obviously (sigh) sounded different and a particular unit sounded better to many panelists. After we connected the units through an A/B switcher, the difference between the units was not so obvious. I could put some science on this because one can measure and detect the existence of the ABX box, but we then would fall into a shouting match about how significant my measurement is to the overall outcome.

I think that the issue here is that the science is low precision, too many science types do not appreciate this, other science types will openly acknowledge this lack of precision, and the emotional types capitalize on this to discredit science and make wild claims about their infallible auditory abilities. Yet, these wild auditory ability claims can be embarrassed with very simple experiments.

It's a mess, but in the end my dog really is really better than your dog -- get over it.

I've been a party to lots of A/B showdowns and the results were mostly "my dog is better than your dog, period". No one wins and everyone goes home mad.

I would agree with this - and it's usually the case that the "experimenters" (if you can call them that) don't understand what they are really testing nor the scientific method.

Even outside of the testing issues encountered in sensory comparison tests and absent the distortion introduced by the equipment you mention, if the "experimenters" were actually using science, they could only make a hypothesis and develop proof for it, but never completely discount the null hypothesis.

Basically, in layman's terms - a positive result establishes, to a certain level of confidence, the reality of different perceptible acoustic differences, a null result, or a failure to reliably detect a difference, does not indicate the nonexistence of that difference.

The simple fact is that sensory comparisons based on memory are 100% fallible and, as such, are not good science. This isn't to say that all the hi-res acoustics and fancy mumbo-jumbo are actually audibly better -- it's to say that testing it is virtually impossible.

Leaning on these tests and relying on them - as many objectors seem to - is truly as much a laughable concept as people that stamp their feet about how much of an audible difference there is between them.

There is a reason eye-witness testimony is viewed incredulously in trials. Any prosecutor, worth his weight in salt, knows that even the most honest and sincere eyewitness testimony is not always credible. Observing an event is not the same as actually remembering the event accurately. Human memory is not a Tivo with the ability to replay with accuracy.

I'll ask again, show me the peer reviewed scienctific study that discounts properly conducted double-blind audio tests as "bad science." there is none. They are accepted as valid science. And nobody is saying the null argument is proven. That is a logical fallacy, and no self-respecting scientist will ever claim a negative is proven. What they are claiming is the differences that may exist cannot be detected by human subjects in properly conducted double-blind tests. So quit putting words in people's mouths. And if your own version if mumbo jumbo is valid science, then link to the studies which show it. :rolleyes:

Basically, in layman's terms - a positive result establishes, to a certain level of confidence, the reality of different perceptible acoustic differences, a null result, or a failure to reliably detect a difference, does not indicate the nonexistence of that difference.


I'll ask again, show me the peer reviewed scienctific study that discounts properly conducted double-blind audio tests as "bad science." there is none. They are accepted as valid science. And nobody is saying the null argument is proven. That is a logical fallacy, and no self-respecting scientist will ever claim a negative is proven.

Guys, I honestly think you are both agreeing with each other without knowing it. I don't want to see this turn into an argument on that basis.

To summarize, I think we (as in myself, jgatie, gtyper, and Buzz) all agree that:

* Double Blind tests (with the assumption they are properly conducted) can specifically prove the case in which there are clear, obvious, and statistically significant differences, but cannot disprove any differences absolutely.

* If there were gross differences between standard res and hires formats they would show up and be conclusively proved in double blind tests. The absence of such results after many years and several tests is enough to conclude that gross (meaning "obvious" or "night and day" to use the words of the hires zealots) differences do not exist.

* The existence of subtle audible differences cannot be completely dis-proven using DBT (or any other proposed scientific methods), but the results are stacked more against them than they are for them. I think all of us probably agree that, in many if not most (or all) cases, the differences between standard res and hires may not be audible.

* Based on hires differences being subtle at best, they will not be audible to most people. I saw a figure somewhere (I don't have a reference) which said that less than 0.05% of people in the UK spend more than £2000 on any room on their audio setup.I personally feel that that is way below the level of spend you need to resolve hires if it is audible.

* The reasons for wanting hires extend beyond pure audibility (for example, the mere presence of superior material, the complexity/inconvenience of down-converting and multiple libraries, the marketing benefits, etc.)

Of course I may be wrong, but I think we are actually all pretty much in agreement with these principles. In which case, the discussion can probably evolve.

I think the big problem of this discussion is that it's all black and white when the reality is grey.

Cheers,

Keith

Majik, I appreciate you trying to moderate, but on these points I believe gtyper does not agree:

* Double Blind tests (with the assumption they are properly conducted) can specifically prove the case in which there are clear, obvious, and statistically significant differences, but cannot disprove any differences absolutely.

* If there were gross differences between standard res and hires formats they would show up and be conclusively proved in double blind tests.

He said as such earlier in the thread when he discounted DBT's that were conducted on compression codes which showed people were indeed able to statistically hear the difference (emphasis mine):

For the record, I come from a science research background - and my opinion is that a blind test in this regard (regardless of the results) is not necessarily good science.

. . .

I did a little digging and found a report wherein the difference between 44.1khz and 96khz was statistically significant. People could, based on the results, find a difference. But, even though I might agree with the assertion, I disagree with the testing methodology and thus believe it not capable of being able to prove anything. I find the test incorrect and I'm sure if you looked at the results - the actual conclusion would be inconclusive. I would argue, that the correlation between the data would not be grouped in a method that would lead one to the original assertion.

It's the nature of using the wrong test methodology. Unfortunately, we don't have a different methodology for comparison other that actually looking at the raw physics and making assumptions on whether or not people can hear the difference.

. . .



And needless to say, on his points above, I most assuredly disagree. Which is why I ask for peer-reviewed articles that back up his assertion that DBT's of audio are not "good science." So far he's produced none.

Majik, I appreciate you trying to moderate, but on these points I believe gtyper does not agree

Actually, I do. Note Majik said:

"[i]Double Blind tests (with the assumption they are properly conducted) can specifically prove the case in which there are clear, obvious, and statistically significant differences, but cannot disprove any differences absolutely.[/i]"

I agree whole-heartedly with this statement. My point, and perhaps I haven't been fully clear, I do not agree with ABX testing methodology for subtle differences. There have been numerous instances where Double Blind AB testing has yielded consistent results in subtle differences but the same test data in an ABX environment fail to produce statistically significant data. This implies a flaw inherent in the test itself and/or a weakness in our sense memory.

Any psychologist will tell you that human memory, especially short term, is a very weak point. This is known and accepted, whether or not you wish to assign any credence to it is another matter.

What I am arguing is that the ABX testing methodology is unsound and poor science for what you're trying to prove in this instance.

I question why we have come to accept ABX methodology for audio to be the gold standard, yet rarely do other sense comparisons use? Why aren't we reviewing the testing methodology to ensure it is actually giving us proper results?

Let's use the Double Blind ABX test for something we know is different. Take shades of the same primary color group. Test it similar to the ABX audio test with the null hypothesis being "One cannot perceive the difference between shades." I am almost certain, in an ABX situation, the null hypothesis would not be disproven. Yet, we know when the colors are placed touching each other (direct visual comparison), the perceptible difference is there thus proving the null untrue. This would point to a flaw in the methodology.

Human's ability to rely on memory for sensory comparison is simply weak. Sound being probably the weakest. Trying to test subtle differences without direct comparison is flawed in the way it is being handled.

Which is why I ask for peer-reviewed articles that back up his assertion that DBT's of audio are not "good science."

I promise you that I spent a good deal of time coming up with a reply. Unfortunately, the system logged me out and it was lost to the ether. I did not cite peer-reviewed articles, but did mention some studies I knew of that showed certain aspects of what I'm talking about but mentioned I was far too lazy to look them up. Now I'm doubly lazy, and half cross-eyed.

The reasons for wanting hires extend beyond pure audibility (for example, the mere presence of superior material, the complexity/inconvenience of down-converting and multiple libraries, the marketing benefits, etc.)

At the end of the day, that's all I care about. I don't care about the difference. I'm not even convinced there is one, as I've stated numerous times.

I simply take offense to the ABX test used, in my opinion, improperly.

I promise you that I spent a good deal of time coming up with a reply. Unfortunately, the system logged me out and it was lost to the ether. I did not cite peer-reviewed articles, but did mention some studies I knew of that showed certain aspects of what I'm talking about but mentioned I was far too lazy to look them up. Now I'm doubly lazy, and half cross-eyed.

Sorry, but without peer-reviewed studies, this is all audiophile hogwash, meant to cloud the results of accepted science. If you haven't got peer-reviewed science, then it's not science. The study I have cited has been peer-reviewed and no one, despite all the money Naim and others stand to lose, have stepped up to debunk it in a scientific journal.

And as far as you agreeing with the tests for gross differences, you also derided a study that showed gross differences between compression techniques using double-blind methods. Forgive me if I find your claims to accept them for non-subtle differences a backslide from your earlier statements.

Oh, and I would appreciate it if you would apologize for saying I or any other proponents of double-blind audio tests implied in any way that negative results proved no differences, which is antithetical to the scientific method. I know I never did, and I don't appreciate you putting words in my mouth in an attempt to state I and others like me are just as bad as the audiophile snakeoil fans.