[PDVJAM]

Well, as I said before, I did several ABX tests myself and these album (or type of music) is the only one where I can say "there is a difference".
Primarily I listen to metal and other heavy genres and in ABX tests can tell the difference between redbook and <200kbps mp3, but if it's 200+ then I'm usually can't hear any difference

 

[RRod]

  Well, as I said before, I did several ABX tests myself and these album (or type of music) is the only one where I can say "there is a difference".
Primarily I listen to metal and other heavy genres and in ABX tests can tell the difference between redbook and <200kbps mp3, but if it's 200+ then I'm usually can't hear any difference

 
Just made my own mp3 then converted back to original sample/bit spec; can't hear anything different in any reverb. Any specific time you latched on to?

 

[PDVJAM]

   
Just made my own mp3 then converted back to original sample/bit spec; can't hear anything different in any reverb. Any specific time you latched on to?

~1:52 - 2:13

 

[RRod]

 
~1:52 - 2:13

 
Yeah, nothing. Here's that section cut out from the two files I'm ABXing:
https://drive.google.com/file/d/0BwmVtb5IwniEczQ1Q2RNTmFnRmc/view?usp=sharing
 
If you can ABX them successfully, then I guess I just have bad/lazy ears, which is of course a real possibility ^_^ Thanks for actually giving details; most people just say "it's just a general feeling you get."

 

[PDVJAM]

Quick try:
 
foo_abx 2.0 report
foobar2000 v1.3.8
2015-04-11 10:59:18
File A: a0.flac
SHA1: 396d5f08ac6fb5590e21b499e87e96d12f8e96dc
File B: a1.flac
SHA1: a6095aeeee0c3f463a3cc2a336ac30122345dbb2
Output:
WASAPI (push) : Speakers (ASUS Xonar Essence ST Audio Device), 24-bit
Crossfading: NO
10:59:18 : Test started.
10:59:27 : 01/01
10:59:47 : 02/02
10:59:53 : 03/03
11:00:03 : 03/04
11:00:07 : 04/05
11:00:12 : 05/06
11:00:25 : 06/07
11:00:32 : 07/08
11:00:32 : Test finished.
 ----------
Total: 7/8
Probability that you were guessing: 3.5%
 -- signature --
74753227474bc55c2c9a6b062eed6808e77d3963
 
With HD650. Will repeat at evening/tomorrow, have to go now.
 
But the truth is: the difference is not like "holy ****, it's obvious like night and day!", but "well, it can be heard, something like voice reverberations in several moments, but it is very, very subtle".

 

[RRod]

 
But the truth is: the difference is not like "holy ****, it's obvious like night and day!", but "well, it can be heard, something like voice reverberations in several moments, but it is very, very subtle".

 
Well, now that you say that you don't have any contention from me ^_^
 
I guess I misspoke a bit: for the person doing the ABX, the difference is usually "obvious". At least when I pass an ABX, I'd have no trouble maintaining a significant percentage even for high trial counts. So even if the difference is subtle, passing the test isn't. I'm willing to believe people can eek out differences in a 320k mp3 file, even if I can't. After all, what's being dropped is content *in the audible range*. If we look at the differences between the original and an uncompressed downsample vs. the original and the mp3, the mp3 differences are pretty much a lion's roar within the audible band! Still most people, like myself, can't hear the difference, probably due to old and beat up ears.

 

[PDVJAM]

Why the difference must necessarily be obvious ? You do not consider an option when there is a difference , but very elusive ? This is such a case - there is a difference, but in can be heard only in several moment of a song (by me).

 

[jcx]

it certainly feels more comfortable when you perceive an obvious difference
 
but it turns out that blind testing with adequate controls can have the statistics show discrimination when the participants can't articulate what or why they choose even to themselves
 
and while the things that are well proven to fine resolution in controlled listening tests seem to be better discerned in quick switching - there's no hard rule that listening time has to be short
 
 
of course in audio the more common outcome is the "night and day" differences sworn to by golden eared audiophiles can't be discerned in controlled listening tests

 

[RRod]

  it certainly feels more comfortable when you perceive an obvious difference
 
but it turns out that blind testing with adequate controls can have the statistics show discrimination when the participants can't articulate what or why they choose even to themselves
 
and while the things that are well proven to fine resolution in controlled listening tests seem to be better discerned in quick switching - there's no hard rule that listening time has to be short
 
 
of course in audio the more common outcome is the "night and day" differences sworn to by golden eared audiophiles can't be discerned in controlled listening tests

 
To me if you can consistently pass ABX tests between two files, then you can hear obvious differences, even if you can't exactly clip out the exact sounds where you hear the difference or explain them in words. People often mention the no-time requirement, but I've yet to see someone get positive results when they sit there for hours listening. Take the current example: PDVJAM doesn't seem to be agonizing over each trial for hours; seems like he sits down, closes his eyes, listens closely, and hears the difference (esp. now that his ears know what to listen for). But he can confirm that for us. Regardless of semantic disagreement, though, I think we all agree that these instant "night-and-day" differences claimed by golden-ears should be easily confirmed by testing, testing they always refuse to do.
 
My only other question in the current situation is if foobar lets you stop the test before the original set number of trials has been reached and still give output for probabilities.

 

[PDVJAM]

   
My only other question in the current situation is if foobar lets you stop the test before the original set number of trials has been reached and still give output for probabilities.

 
Well, I did all these tests with opened eyes and as you can see, for every attempt I need several seconds.
 
PS: foobar2000 will not show you any results if you quit before you passed all tests... So for me optimal is 10 tests. More than 10 - and I'm starting to hate this music

 

[RRod]

   
Well, I did all these tests with opened eyes and as you can see, for every attempt I need several seconds.
 
PS: foobar2000 will not show you any results if you quit before you passed all tests... So for me optimal is 10 tests. More than 10 - and I'm starting to hate this music

 
Several seconds is fine, I just wonder if anyone ever does these things and actually needs minutes for each successful test. And yeah, it can be dangerous to ABX tracks one really likes ^_^

 

[PDVJAM]

   
Several seconds is fine, I just wonder if anyone ever does these things and actually needs minutes for each successful test. And yeah, it can be dangerous to ABX tracks one really likes ^_^

 
Well, in my case and with this track the difference is small and can be noticed only with very quick switching. I think there are 2 available differences - obvious, and it doesn't need for long a listening coz is obvious and subtle, and it doesn't need long listening as well, coz subtle difference can be heard only with quick switching.
Things like "i need to live with this setup for a week to hear a difference" I perceive as a total BS.

 

[RRod]

   
Well, in my case and with this track the difference is small and can be noticed only with very quick switching. I think there are 2 available differences - obvious, and it doesn't need for long a listening coz is obvious and subtle, and it doesn't need long listening as well, coz subtle difference can be heard only with quick switching.
Things like "i need to live with this setup for a week to hear a difference" I perceive as a total BS.

 
I agree with you here. I mean, maybe I could see long-term listening work if you literally hard-coded the music into your brain on a particular setup at a particular volume in a particular ambience, but no one ever does that. And I like those definitions of obvious vs. subtle in passing ABX tests.

 

[castleofargh]

any talk about needing a long time to notice differences between 2 pieces is really 187.18% BS. there isn't one argument in the world that would support this. anybody with enough understanding of the human memory knows that.
if the question was, tell us what are the differences? or how many are there? then more time might really help, but in abx the question is "is there any difference?" and to answer that the very best a human can do is having both pieces as close as possible in space and time. and how fast the switch is done has been showed to be the most important part of all, be it for visual or for acoustic observation. there are countless visual stuff done like that where just adding half a second black screen between 2 images makes people much more susceptible to miss any change that occurred in the picture. while they pretty much always notice a change when its instantaneous(again they might not know what changed, but they will know that something moved). ABX is so good for human because of how seamless the switch can be done and how fast we can jump from one to the other.
 
"ok I've seen the murder, now the best way to identify him is to live with 10 suspects for a month and only then will I be able to best tell if one did it!" how stupid would it be to say something like that? but in audio people just have no shame, or really no understanding, and I guess mostly both, so they claim stuff like that all the time.
 
our aptitude to adapt to our environment to survive is misunderstood as "burn in" by most people.
the way we only remember the most noticeable things and exaggerate them as a mean to remember them better when stored in long term memory, it's misunderstood as some ability to tell things apart better after more time.
our inherent habit to use logic shortcuts and look for patterns that agree with our first idea, is seen as validation that we're right and don't need an ABX. or even that ABX is wrong because it didn't agree with us when things usually do.
 
just understanding humans a little is enough to show where all the audiophile nonsense is coming from. and most of the time it has nothing to do with audio. they're just humans trusting everything their brain is telling them. how amazing a magician must be to those guys.

 

[edstrelow]

I think many people in this forum expect more from science than it can actually deliver.
 
It is very hard to get meaningful measurements about equipment that will tell you something definitive about whether the equipment is good or not.  Sometimes this is because the issues are more complex than most people assume and sometimes  because we may not be able to pinpoint what it is we need to measure which will predict what listeners will hear and like.
 
it is also naive to expect much in the way of good data from most human testing in this field, and I say this as a one-time experimental psychologist who did this type of work for many years. As has been noted above,  comparisons of auditory samples such as the sound, from  two different pieces of equipment are limited by the fact that we cannot simultaneously compare the 2 different  sounds, we must rely on memory.  (The ABX method does not solve this problem but may give a marginal boost to memeory)
 
The same is not true in vision where you can put 2 or more stimuli in front of an observer simultaneously.  Thus auditory testing of comparisons of 2 stimuli is  statistically noisy and flawed in the direction of not finding a difference which might be detectable if a better test could be devised.  The situation gets even worse when you try to get groups of persons to detect differences, because you you are adding in unknown variability of the judgment processes of multiple persons.
 
There are other techniques which could be applied, for example what is sometimes called "the method of adjustment."   A simple example many of us use is found in the various photo touch-up programs. We may adjust a parameter, such as contrast, up or down until we see something we like.   Often I don't think I could reliably tell the beginning and end product apart if I had them in front of me  in succession as would happen with auditory stimuli, or possibly even simultaneously.  But that wouldn't mean that there was no detectable difference.
 
To me the best testing still is to take a familiar recording and if you can't initially notice anything better or worse with different equipment you probably won't care much one way or the other even if there is actually a difference  there.  Then maybe you can refine your impressions with simple AB comparisons.