Some questions about settings (sample rate, latency, etc.) for an external USB DAC running off of ASIO
Aug 6, 2015 at 5:34 PM Post #121 of 138
actually you may spend some time just listening for A vs B to try and identify a recognizable feature where you hear a difference
 
although "forced choice" has been reported to sometimes give statistical positive result even when the test subject reports confusion/no reliable conscious discrimination 
 
Aug 6, 2015 at 6:06 PM Post #122 of 138
  actually you may spend some time just listening for A vs B to try and identify a recognizable feature where you hear a difference
 
although "forced choice" has been reported to sometimes give statistical positive result even when the test subject reports confusion/no reliable conscious discrimination 

 
This is one of several relatively well known flaws in what at first glance seem to be reasonable and effective test protocols; many of these can be summed up as "what the test subjects DO may not be the same as what they SAY".
 
Many test subjects prefer to avoid submitting a reply unless they're certain. So, for example, you may find that most test subjects report that they don't prefer either Brand X or Brand Y. However, if you leave them in the room with both, you find that the majority finish off the Brand X, while most leave the remainder of Brand Y untouched. This is a very interesting result, and has all sort of ramifications... especially in marketing and advertising.
 
If those same subjects walk into a supermarket, and have the option of buying Brand X or Brand Y, will they actually purchase equal amounts of either (because they really don't believe there's a difference), or will they choose mostly Brand X (because whatever unconscious bias caused them to finish Brand X more often in the test serves as "a tie breaker" when they find themselves consciously unable to choose)? 
 
If it turns out that the majority of participants SAY that they don't have a preference, but in fact choose Brand X when forced to choose, then you'll probably want to figure out how to get the actual participants into the store. (Because, if they walk into the store, most of them will buy Brand X but, if they send a friend to shop for them, they'll probably tell that friend that they don't care which he or she buys. In which case you need to know which their friend prefers. However, if you're planning a fancy ad campaign to "program" people to buy Brand X, then you'd better make sure you target your ads to the folks who do the shopping.)
 
Whole marketing studies have been devoted to finding the answer to that question in specific instances
very_evil_smiley.gif

 
Aug 6, 2015 at 6:27 PM Post #123 of 138
Before there is even a point of doing an ABX test, you need to perceive a difference between A and B. You should have a good sense of what differences you perceive between the two, in order to look for those differences during the ABX test.
 
Aug 7, 2015 at 4:37 AM Post #124 of 138
   
This is one of several relatively well known flaws in what at first glance seem to be reasonable and effective test protocols; many of these can be summed up as "what the test subjects DO may not be the same as what they SAY"

 
I suspect that this affects high end audio quite profoundly, as many people will not argue with a lot of the craziness, but they won't buy it, either.  And we have plenty of evidence that even people who buy into it, have serious doubts and will probably not repeat the mistake.
 
Aug 7, 2015 at 4:39 AM Post #125 of 138
  Before there is even a point of doing an ABX test, you need to perceive a difference between A and B. You should have a good sense of what differences you perceive between the two, in order to look for those differences during the ABX test.

 
There is plenty of evidence that people can obtain positive results from ABX tests in which they never consciously hear a difference.
 
Been there, done that.
 
What is true is that hearing a difference is a logical prerequisite for having a preference.
 
Aug 7, 2015 at 4:45 AM Post #126 of 138
   
Not at all. The only bias that I will (cheerfully) admit to is that I'm extremely skeptical when someone claims that there is an audible difference,

 
What you are willing to admit to is meaningless, logically and scientifically speaking. It is based on the fallacy that people can possibly always know what all of their biaess are. Those individuals may exist but they are not human! :wink:
 
Aug 7, 2015 at 5:00 AM Post #127 of 138
   
Agreed.
 
I also noticed that the Foobar ABX test plugin allows the listener to switch back and forth between A, B, and X as many times as they like and in any order. This eliminates the main objection that I have to what I would call "formal ABX testing". (In what I would call formal ABX testing, the listener hears A, then B, then X - and is then asked to make a value judgement about which reference sample matches X, which requires some level of "acoustic memory" of both A and B. With the Foobar plugin, one may instead choose to go directly back and forth between A and X several times, and decide if they are identical or not, then go directly back and forth between B and X, and decide if they are identical or not. This would seem to me to require less acoustic memory, and so to be a more direct and sensitive indicator of "similarity".) 

 
What you are talking about above is the fact that there are two different tests called ABX test in the scientific literature.
 
One was first described in the scientific literature ca. by 1950 Munson, W.A. and M.D. Gardner. "Loudness patterns - A new approach". Journal of the Acoustical Society of America, 22:177–190, 1950.   it is a test that was administered by a machine that would  play A, then B, then X - and the listener is then to make a forced response identifying X as either A or B.
 
This is far from being an interactive test, but it still makes sense for certain kinds of testing to this day.  it basically limits the ability of the listener to learn about the listening task and optimize his performance by means of repetatition, which equates to things like speech intelligibility in a lecture hall.
 
Independently in 1977 an interactive test was developed where the listening test was administered by a machine that would  play A,  B, and X any number of times, switched at any time under the full control of the listener, until the listener was as comfortable as possible with his identification of X.  Clark, D. High-Resolution Subjective Testing Using a Double-Blind Comparator. JAES Volume 30 Issue 5 pp. 330-338; May 1982.
 
This is a highly interactive test, and it makes sense for modelling how people listen to music for pleasure .  It does not limit the ability of the listener to learn about the listening task and enables and even encourages him to optimize his performance by means of learning and repetition.
 
Both kinds of tests are formalized by means of publication in peer-reviewed publications long ago, and have been discussed informally, in many conference papers and peer-reviewed publications for many decades. There is no excuse for a person who purports to speak authoritatively about them to be confused about them at this time, but it happens all the time.
 
Aug 7, 2015 at 10:57 AM Post #128 of 138
   
What you are willing to admit to is meaningless, logically and scientifically speaking. It is based on the fallacy that people can possibly always know what all of their biaess are.Those individuals may exist but they are not human! :wink:

 
Actually it is simply a statement of my "assumptions" - which is not saying that it "proves" anything. I believe that any difference that can be heard can be measured (which is not to say that we necessarily are measuring the right things to detect them). However, if someone claims to hear a difference, but cannot produce any measurements that show a difference, then I consider their claim to be less likely to have merit than if they claim to hear a difference, and we also know that measurable differences do exist, so the only question is whether the differences (which we already do know exist) are audible. My (personal) "screening process" considers that claims of audible differences that are associated with documented measured differences are much more likely to be true that those in which there is no measured difference associated with the claim.
 
It's the same as my saying that I would be more likely to believe someone who said they saw a horse in their front yard than someone who said they saw a unicorn in their front yard - because we all know that at least horses do in fact exist (and can reasonably wander into front yards). Therefore, while both are in fact not technically impossible, I might get out my camera and go for a drive to perchance get a few shots of the horse, but I probably wouldn't bother to make the trip on the very-off-chance of getting a picture of a unicorn. (I could be wrong in either case, but, statistically, I'm a lot less likely to be wasting my time if I follow up on "horse stories" and ignore "unicorn tales".)
 
As for biases - I disagree with you there. I agree that we, as individuals, do a relatively poor job of accurately gauging or observing our own biases directly. However, lots of studies have been done about individual biases held by certain individuals and groups, and on how biases are formed, and on how biases affect our "perception". Therefore, I do have a pretty good idea what biases I (and you) are likely to have - based on generalizations about human nature. For example, since we've been "taking opposite sides" on a certain discussion, we are both probably biased to find evidence that supports our point of view "more believable", and evidence that opposes it "somewhat less credible". Another common and well-known bias is that most Americans have a bias towards assuming that more expensive products are in fact better ("you get what you pay for"), so, all else being equal, they tend to be biased towards believing that "expensive wine tastes better" and "expensive equipment sounds better"; while, in many other cultures, this bias is not as prevalent. Studies have been done; it almost always works that way; and I have no reason to suspect that either of us is immune to it.
 
Now, speaking for myself, it has been my experience that differences which are claimed to be audible, but have no "numbers" whatsoever to suggest that a difference exists, have consistently turned out to be untrue, and that, when measurable differences do exists, they at least sometimes do turn out to be audible - which does in fact bias me towards being willing to spend effort whenever there seem to be numbers that "might reasonably seem to account for" a claimed audible difference - or, if I'm feeling especially lazy, "to give them the benefit of the doubt. (I think it's only fair for anyone who is trying to decide how much value to place on my opinion to understand that bias... especially since, as far as I know, that is not necessarily a bias that is common to the vast majority of humans.)
 
Aug 7, 2015 at 12:42 PM Post #129 of 138
  There is plenty of evidence that people can obtain positive results from ABX tests in which they never consciously hear a difference.
 
Been there, done that.
 
What is true is that hearing a difference is a logical prerequisite for having a preference.

 
Well...that's pretty interesting, actually.
 
But I just meant that if I don't perceive a difference between two audio files, I wouldn't have much reason to do an ABX test. In most contexts on this site, people do an ABX test to determine whether a difference they are perceiving is genuine or placebo.
 
Aug 30, 2015 at 12:25 AM Post #130 of 138
Hi. This probably isn't the proper forum for this question, but I can't find a proper forum and this is the closest.  Maybe someone can help me.  Is there a rule of thumb (preferably a rule based on measurement/science) for what level to set the volume at the Laptop when using a Laptop as a source for a DAC/Headphone Amp? The higher the volume at the laptop, the more sensitive the volume at the amp will be, for sure, but what I really want to know is what will produce the best sound quality?  50%?    70%?  Typically, when there is a "line out" vs. "headphone amp" the "line out is considerably quieter than the headphone out at max.  I'm sure there is a rule of thumb.  Anyone?
 
Aug 30, 2015 at 1:17 AM Post #131 of 138
  Hi. This probably isn't the proper forum for this question, but I can't find a proper forum and this is the closest.  Maybe someone can help me.  Is there a rule of thumb (preferably a rule based on measurement/science) for what level to set the volume at the Laptop when using a Laptop as a source for a DAC/Headphone Amp? The higher the volume at the laptop, the more sensitive the volume at the amp will be, for sure, but what I really want to know is what will produce the best sound quality?  50%?    70%?  Typically, when there is a "line out" vs. "headphone amp" the "line out is considerably quieter than the headphone out at max.  I'm sure there is a rule of thumb.  Anyone?


set the computer as high as you can. then if you end up clipping because of some DSP or EQ you're using, turn it down until it doesn't ^_^.
same thing if the amp ends up with an audible channel imbalance, or simply if you have a hard time setting the volume level you like. ideally you would change the gain setting on the amp, but all amps don't offer that choice(usually happens to people who bought 3W amp because some dudes told them it was the best, but they use it to drive a momentum and some IEMs). so if you have no other choice, then setting the computer output to 24bit and lowering the volume setting a little might just be the practical answer.
 
so there you have your rule of thumb, if you lower the volume on the computer, with most DACs that will result in losing that much bits in the quieter part of the track, so while it's not audible until you really abuse it, it's still a degradation of the signal. just don't go making your life impossible for a stupid principle. that would be my honest advice.
biggrin.gif
 
 
Aug 30, 2015 at 4:38 AM Post #132 of 138
   
Actually it is simply a statement of my "assumptions" - which is not saying that it "proves" anything. 

The above statement seems to show a complete misunderstanding of Science in the 21st century. In modern science there is no such thing as proof, just reliable and unreliable evidence.
 
It is well known to people who actually study and understand modern science that all findings of science are provisional until more reliable evidence is found.
 
Thus, we can operate constructively in a world where Newton's findings about the laws of motion describe 99.999+% of all of our experiences well enough to build most SOTA systems, but we also know that Einstein showed how they were false almost 100 years ago. In a few exceptional cases Einstein's findings are needed to even be in the right zip code.
 
Aug 30, 2015 at 4:47 AM Post #133 of 138
 
set the computer as high as you can. then if you end up clipping because of some DSP or EQ you're using, turn it down until it doesn't ^_^.
same thing if the amp ends up with an audible channel imbalance, or simply if you have a hard time setting the volume level you like. ideally you would change the gain setting on the amp, but all amps don't offer that choice(usually happens to people who bought 3W amp because some dudes told them it was the best, but they use it to drive a momentum and some IEMs). so if you have no other choice, then setting the computer output to 24bit and lowering the volume setting a little might just be the practical answer.
 
so there you have your rule of thumb, if you lower the volume on the computer, with most DACs that will result in losing that much bits in the quieter part of the track, so while it's not audible until you really abuse it, it's still a degradation of the signal. just don't go making your life impossible for a stupid principle. that would be my honest advice.
biggrin.gif
 

 
Actually, lowering the volume does not degrade a digital output any more than lowering the volume of an analog output. They both drop the signal into the real world morass of background noise.  
 
Furthermore, the strongest effect of dropping the volume is psychoacoustic - it is in your ear/brain system. Your ability to reliably detect small differences is usually optimal for SPLs on the order of 80-85 dB SPL plus minus 5 or so dB. However there are a number of common hearing maladies that can affect this. 
 
Nevertheless the basic advice is good - run the level of the laptop as high as you can without audible distortion and use the gain control on the headphone amp to obtain fine adjustments of listening level. This optimizes what is known as gain staging. Both noise and distortion are then optimized for your application. 
 
Aug 30, 2015 at 4:52 AM Post #134 of 138
   
Well...that's pretty interesting, actually.
 
But I just meant that if I don't perceive a difference between two audio files, I wouldn't have much reason to do an ABX test. In most contexts on this site, people do an ABX test to determine whether a difference they are perceiving is genuine or placebo.

 
This is all true, fine and good, but just because a lot of people use ABX tests to distinguish between illusions and reliable perceptions doesn't mean that is all they are good for.
 
The various common biases that many fill their lives with by relying on sighted evaluations do a lot more damage than just creating or reinforcing illusions.
 
They also make it impossible to obtain reliable knowledge about just about everything that is heard. 
 
Aug 30, 2015 at 1:47 PM Post #135 of 138
  The above statement seems to show a complete misunderstanding of Science in the 21st century. In modern science there is no such thing as proof, just reliable and unreliable evidence.
 
It is well known to people who actually study and understand modern science that all findings of science are provisional until more reliable evidence is found.
 
Thus, we can operate constructively in a world where Newton's findings about the laws of motion describe 99.999+% of all of our experiences well enough to build most SOTA systems, but we also know that Einstein showed how they were false almost 100 years ago. In a few exceptional cases Einstein's findings are needed to even be in the right zip code.


YUP!  General Relativity, to account for the curvature of space by the earth's gravity, is needed to achieve accuracy to within more than a few tens of meters for GPS systems :wink:
 
There is no TRUE "proof" in any field besides Mathematics.  But Mathematics is not a branch science, it is simply USED more than anything else in science.  And the Scientific-Method itself is precisely as you say. . .it is a method of showing that a particular hypothesized formula, equation, or some other such thing is a good enough approximation to the true behavior of a system to allow us to make reliable, useful predictions with it to within the tolerance of our measuring equipment or acceptable margin-of-error for our applications.

Anyway guys, it's neat to see that my thread has been revived from the Head-Fi Underworld to walk among the living once again
biggrin.gif

 

Users who are viewing this thread

Back
Top