They aren't. He just doesn't know that they aren't.
Enviado de meu GT-I9505 usando Tapatalk
It's difficult to speak of this accurately.
The links are the framework of how our mind works, it's basically a platform which potentially could lead to that slow switching could reveal more information than fast switching, in some instances.
We need memory to remember what A and B sound like, as linked, sonic "noise memory" is accurate and lasts for weeks, it works by forming new cells, the signal is pretty much stored like a document in our mind, that is evidence that acoustic memory is not strictly short, in other words, sound is not "new" every single time we hear it.
A sound signal is sometimes matched to a long-lasting and accurate memory we have of countless sound signals.
The hypothesis I've seen that fast switching is absolutely necessary, states we have no such long-lasting file system, thus, that statement is incorrect.
If we entertain that we can remember what the "voice" of an amplifier sounds like for weeks or years, after the necessary exposure to it, that still doesn't explain why the difference would disappear in a fast switch, right?
Then we delve into theory, i.e. what is our pre-attentive processing deleting, what is altered and what is passing through unaltered.
The pre-attentive is avoiding unnecessary information, if we heard everything we'd be "schizophrenic" so to speak, less efficient.
In fact some people I think have a condition like that, they would be "necessary" to show what is deleted during an ABX trial, if anything is, to us normal humans.
I think that should summarize the framework.
Last edited:
Others will know more about this than me, but I suspect that it is not a "hypothesis" but a demonstrated experimental fact that the slower the switching the harder it is to distinguish between genuinely different sounds.Kastor L said:
We have a good long term memory for tunes. We have a poor long term memory for sounds.
we have a near-perfect memory for random noise, like kinds of static noise I think, it doesn't need to be music
Edit - It would be helpful if the files were available that they used in that test, it's not entirely clear, but the cell formation part is the most interesting aspect I took from it anyway
Edit - It would be helpful if the files were available that they used in that test, it's not entirely clear, but the cell formation part is the most interesting aspect I took from it anyway
Last edited:
I'm wondering if NwAvGuy, Winer and the likes of them arrived at that with intuition or with an actual study
I suspect that our "volume memory" is pretty horrid
Then there's studies on pitch, you know, perfect pitch and all that
You could then, if you want to, incorrectly, refer to the volume and pitch studies as encompassing all acoustics
Here is what the Wiki states at least
"For acoustic pre-attentive processing, the temporal cortex was believed to be the main site of activation, however, recent evidence has indicated involvement of the frontal cortex as well.
The frontal cortex is predominantly associated with attentional processing, but it may also be involved in pre-attentive processing of complex and/or salient acoustic stimuli.
For example, detecting slight variations in complex musical patterns has been shown to activate the right ventromedial prefrontal cortex.
It has been shown that in acoustic pre-attentive processing there is some degree of lateralization.
The left hemisphere responds more to temporal acoustic information whereas the right hemisphere responds to the frequency of auditory information.
Also, there is lateralization in the perception of speech which is left hemisphere dominant for pre-attentive processing"
Last edited:
I find the process of picking differences very straightforward - most seem to think that you should be listening for the better version; wrong, wrong, wrong, IMO. What needs to happen is that you listen for the worst bit in the reproduction, the little tidbit where the sound is really crappy, unrealistic - which may only last for a second or two. From then on you completely ignore all the rest of the track, it's irrelevant for what you're doing next - you focus in on precisely how ugly that poorly rendered bit of sound is, you mentally turn it into a type of sound effect, so it now has a shape and texture to it.
Unless the ABX setup is pretty bad - and unfortunately foobar's playback fits into that category to a large degree - the A and the B will create two very distinct sound effects from that sample of track you zoomed in on. And that's all you need to use for the distinguishing procedure, the rest of the music is mere filler - and should be disregarded completely.
Unless the ABX setup is pretty bad - and unfortunately foobar's playback fits into that category to a large degree - the A and the B will create two very distinct sound effects from that sample of track you zoomed in on. And that's all you need to use for the distinguishing procedure, the rest of the music is mere filler - and should be disregarded completely.
These Turing devices seem to have evolved a good way to appear to win any argument. First they state something at odds with accepted science. Then in support of that they make an assertion that the accepted science is just a hypothesis. The next step is usually to demand proof that the hypothesis is actually true. This 'proof' may be hard to find at short notice, and often only appears in peer-reviewed journals which are relatively inaccessible to the general public. It may be reported in the standard textbooks, though.
So we get a bifurcation: if evidence of the hypothesis cannot readily be produced they conclude that this is proof that they were right. They of course hope that we won't notice that they are using absence of evidence for something as being equivalent to evidence against it - a logical fallacy.
If evidence can be produced then they can either dispute it (e.g. by saying that it belongs to a different 'paradigm') or they can simply fail to understand it or engage with it. This is a bit like someone who has never done calculus at all telling someone else that they have solved a differential equation in the wrong way - any correct argument necessarily goes over their head but it is considered rude to point this out. We get a lot of this type of stuff with the Fourier deniers and the circuit theory deniers.
So we get a bifurcation: if evidence of the hypothesis cannot readily be produced they conclude that this is proof that they were right. They of course hope that we won't notice that they are using absence of evidence for something as being equivalent to evidence against it - a logical fallacy.
If evidence can be produced then they can either dispute it (e.g. by saying that it belongs to a different 'paradigm') or they can simply fail to understand it or engage with it. This is a bit like someone who has never done calculus at all telling someone else that they have solved a differential equation in the wrong way - any correct argument necessarily goes over their head but it is considered rude to point this out. We get a lot of this type of stuff with the Fourier deniers and the circuit theory deniers.
Kastor, the level of inflexibility with regard to the "magic" of the DBT method is quite something to behold - I've mentioned a couple of times another technique, which compares somewhat to how I go about things: repeat a sample, A, 5 times, so the ears get into a rhythm of picking up the sense of the clip; then without a break do another 5 samples, but somewhere in there it switches to B, and continues with B - could be the 6th one, or the 10th one; perhaps do a couple more at the end to reinforce the sense of the B sample. Obviously, the test is to always pick the transition point, to satisfy people who love numbers, 
.
Of course, not a single person commented on this idea ...
.Of course, not a single person commented on this idea ...
The sector of blind listening per se isn't inflexible at all.
I'm quite sure we are all be on the same page there at least, blind listening itself, in some kind of pure concept, isn't flawed at all.
The only reasons I can think of are truly imagined differences and some kind of stress or mental performance issues.
Like, I imagine that this and that sound different, much like someone imagines that they can feel Feng Shui energy in their house, they just like that world-view.
Naturally such imagined differences can't be associated if you're not aware of your surroundings.
The above is very rare in audio imho, note that I'm referring to semi-conscious imagination, not expectation effect. It's like "well, deep down I think Feng Shui energy doesn't exist, but I prefer to believe in it". Expectation effect and placebo are different.
However, a few will be passionate about it and defend imagined differences, like "just give us a rest" or "we're the ones buying it, not you" or "you need to understand how the economy works, how the human mind works, everything is illusory anyway", and such.
I feel a bit sorry for the people that hear real differences though and are intimidated by science, so they're led to believe that what is truly real is not. For examples sake, if we entertain that SACD is actually real and someone sells their entire collection due to a few tests, that's not really fair now, is it? =)
The other would be that you focus differently during a blind test, like stress, so it becomes more difficult than sighted.
We really need to know what to listen for and how to listen, to hear the difference at all, it doesn't just "appear" like a distinct light / colour to the human eye which is impossible to not see, we really need to know where to look for the difference.
If anyone here thinks the statement directly above is untrue, then just say so, I can provide some kinds of evidence.
Aside from these two, imagined differences and stress / focus, I can't think of anything flawed in blind listening per se right now.
The first isn't really a flaw in blind listening, it's just in conflict with that view of truly imagined differences.
The flaws arrive in the tests themselves and in the human interaction.
For example, John Curl mentioned a test of cables once where people could actually hear the difference.
According to the story, the test director quickly cancelled the test. It seems like he thought something was "broken" in the test since people could actually hear the difference, which is in conflict with his view or his desired result.
The same happened in this thread, like, the
whatsbestforum has a few positive results in 96 kHz recordings, so, then some people here like Julf prefer to "cancel it", like, "that's just an invalid test", or use strawman talk like "So you defy the laws of Nyquist?", when nothing in that positive result seems to be defying Nyquist at all.
The Meyer and Moran study was using redbook material presented as SACD, it's discussed at sa-cd.net.
The Matrix Hi-Fi blanket test is statistically very invalid.
Various ABX tests use random university
students with no listening experience at all, even if they're musicians you're still heading for null results with random participants that are not familiar with the equipment imho.
Then there is the blind test procedure in theory versus in practice.
In a kind of theory / concept ABX might be the most revealing, but there is no evidence available for that and it's quite likely not the most revealing.
Some people may not like my commentary but science is exhaustive per it's design, so it's up to you.
The statistical trend proves that science is ever-changing anyway.
I'm quite sure we are all be on the same page there at least, blind listening itself, in some kind of pure concept, isn't flawed at all.
The only reasons I can think of are truly imagined differences and some kind of stress or mental performance issues.
Like, I imagine that this and that sound different, much like someone imagines that they can feel Feng Shui energy in their house, they just like that world-view.
Naturally such imagined differences can't be associated if you're not aware of your surroundings.
The above is very rare in audio imho, note that I'm referring to semi-conscious imagination, not expectation effect. It's like "well, deep down I think Feng Shui energy doesn't exist, but I prefer to believe in it". Expectation effect and placebo are different.
However, a few will be passionate about it and defend imagined differences, like "just give us a rest" or "we're the ones buying it, not you" or "you need to understand how the economy works, how the human mind works, everything is illusory anyway", and such.
I feel a bit sorry for the people that hear real differences though and are intimidated by science, so they're led to believe that what is truly real is not. For examples sake, if we entertain that SACD is actually real and someone sells their entire collection due to a few tests, that's not really fair now, is it? =)
The other would be that you focus differently during a blind test, like stress, so it becomes more difficult than sighted.
We really need to know what to listen for and how to listen, to hear the difference at all, it doesn't just "appear" like a distinct light / colour to the human eye which is impossible to not see, we really need to know where to look for the difference.
If anyone here thinks the statement directly above is untrue, then just say so, I can provide some kinds of evidence.
Aside from these two, imagined differences and stress / focus, I can't think of anything flawed in blind listening per se right now.
The first isn't really a flaw in blind listening, it's just in conflict with that view of truly imagined differences.
The flaws arrive in the tests themselves and in the human interaction.
For example, John Curl mentioned a test of cables once where people could actually hear the difference.
According to the story, the test director quickly cancelled the test. It seems like he thought something was "broken" in the test since people could actually hear the difference, which is in conflict with his view or his desired result.
The same happened in this thread, like, the
whatsbestforum has a few positive results in 96 kHz recordings, so, then some people here like Julf prefer to "cancel it", like, "that's just an invalid test", or use strawman talk like "So you defy the laws of Nyquist?", when nothing in that positive result seems to be defying Nyquist at all.
The Meyer and Moran study was using redbook material presented as SACD, it's discussed at sa-cd.net.
The Matrix Hi-Fi blanket test is statistically very invalid.
Various ABX tests use random university
students with no listening experience at all, even if they're musicians you're still heading for null results with random participants that are not familiar with the equipment imho.
Then there is the blind test procedure in theory versus in practice.
In a kind of theory / concept ABX might be the most revealing, but there is no evidence available for that and it's quite likely not the most revealing.
Some people may not like my commentary but science is exhaustive per it's design, so it's up to you.
The statistical trend proves that science is ever-changing anyway.
I think an ABX test allows you to listen to A five or ten times so there is nothing controversial in your view, unless the test protocol doesn't allow for it.
The Foobar protocol doesn't allow for slow switching for example, it's not available there and many people speak like it's not allowed, for mystical or even slightly suscpicious reasons.
In theory pretty much anything is allowed in blind testing, you can sit on a camel if that helps.
Long pause switching is still switching though.
I think "identify X" is the real deal in all this, you don't compare, you just identify.
For example, if you have a lemon and a lime, do you really need to run ABX trials and complicate everything?
Don't use non-transparent switches / pathways, just ask them what "X" is twenty times, lemon or lime.
In fact if you switched a lemon to a lime in less than 0.1 seconds the participant might not even notice =)
Last edited:
Foobar is not good enough quality, it fails the "Is the measuring equipment sufficient?" test - so all bets are off after that.
My way is that you create a pattern of sound which you "learn" by its repetition, and then change the pattern at a random instant - the mind is very adept at noticing a slight change in its environment, which doesn't match what it previously experienced - it sticks out like the proverbial sore thumb ...
Foobar's protocol doesn't allow my method, there is too much "rational" input in how you have to drive it - basically, the "thinking" side of the brain has to be kept out of the way as much as possible while doing the test.
My way is that you create a pattern of sound which you "learn" by its repetition, and then change the pattern at a random instant - the mind is very adept at noticing a slight change in its environment, which doesn't match what it previously experienced - it sticks out like the proverbial sore thumb ...
Foobar's protocol doesn't allow my method, there is too much "rational" input in how you have to drive it - basically, the "thinking" side of the brain has to be kept out of the way as much as possible while doing the test.
Last edited:
I don't follow you very well there, but okay.
I don't even care about jitter and DSD time resolution all that much personally, although the whatsbestforum findings are interesting for sure.
I care about DAC chips, DAC units and amplifiers, amplifier IC's in particular.
Capacitors to a lesser extent as well.
I can't run any of the above in a Foobar test, unless I were to believe that I can record a DAC and listen to it in Foobar, via a different DAC, but that is extreme truncation.
I don't even care about jitter and DSD time resolution all that much personally, although the whatsbestforum findings are interesting for sure.
I care about DAC chips, DAC units and amplifiers, amplifier IC's in particular.
Capacitors to a lesser extent as well.
I can't run any of the above in a Foobar test, unless I were to believe that I can record a DAC and listen to it in Foobar, via a different DAC, but that is extreme truncation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.