IPB

Welcome Guest ( Log In | Register )

4 Pages V  « < 2 3 4  
Reply to this topicStart new topic
ABX testing vs. unconscious perception/decision, enjoyment, and so on, Split from “xiphmont—‘There is no point to 24/192’”/TID:93853 (TOS #5)
Woodinville
post Mar 30 2012, 02:06
Post #76





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (icstm @ Mar 29 2012, 02:15) *
QUOTE (Woodinville @ Mar 28 2012, 19:39) *
What you are missing is what can,and what can not, be detected at all on the auditory periphery.
yes I that is exactly right, I am not clear what the response curves are on the auditory periphery. How have we measured this? And by measured, I do not mean asked subjects, I mean where have we done the equivilent of sticking a probe just past the transducer.

Nice set of posts btw, just taken a look.


It's not just "response curves", the SNR of a given inner hair cell (which is the detector) is about 30dB, which is mapped over about 90dB by the compression provided by the outer hair cell.

Of course, an inner hair cell responds only to signals that actually make it move, and the cochlear filter is 70 to 80 Hz in bandwidth at low frequencies, and about 1/4 octave at higher frequencies.

As far as "sticking a probe", it would appear you are asking someone to be seriously wounded in order to determine this in a human being, so let's stop with that foolishness.

As far as "asked subjects", can you explain what your problem with a proper subjective test might be? Such tests are verifiable, repeatable, and falsifiable, and provide the same result over and over and over again, and have from the 1930's to present.

This post has been edited by Woodinville: Mar 30 2012, 02:07


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
Woodinville
post Mar 30 2012, 02:09
Post #77





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (2Bdecided @ Mar 29 2012, 03:10) *
It's been done on cadavers, but a lot of the sensitivity (and all of the adaptability) of the ear dies with its owner.

Cheers,
David.


Quite so, the cochlea (basilar membrane in particular) is an active organ, and the filter shapes change substantially after death. None the less, subjective experiments provide a good measure of the ear's filter bandwidth and SNR at any given instant.


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
icstm
post Mar 30 2012, 10:25
Post #78





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



QUOTE (Woodinville @ Mar 30 2012, 01:06) *
As far as "sticking a probe", it would appear you are asking someone to be seriously wounded in order to determine this in a human being, so let's stop with that foolishness.
I think the answer is simple - namely that when you are testing a system, you want to understand the inputs and ouputs of each module. "asking" someone would be a response to stage 6 or so in my simple (and probably wrong) model. In either case it is too late in the system.

I did not mean actually using a giant probe, I meant using a sensor to test earlier in the chain.
Go to the top of the page
+Quote Post
Woodinville
post Mar 30 2012, 21:09
Post #79





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (icstm @ Mar 30 2012, 02:25) *
QUOTE (Woodinville @ Mar 30 2012, 01:06) *
As far as "sticking a probe", it would appear you are asking someone to be seriously wounded in order to determine this in a human being, so let's stop with that foolishness.
I think the answer is simple - namely that when you are testing a system, you want to understand the inputs and ouputs of each module. "asking" someone would be a response to stage 6 or so in my simple (and probably wrong) model. In either case it is too late in the system.

I did not mean actually using a giant probe, I meant using a sensor to test earlier in the chain.


Sorry, you aren't understanding even the basics of how auditory perception has been shown to work, or how to probe a system response in general.

Look up "signal detection experiment" for starters, how about?


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
icstm
post Apr 2 2012, 10:23
Post #80





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



@Woodinville
You are right, I am not well versed in Signal Detection Theory, and indeed there must be many other theories that given your background in this space, you will be able to educate me and others about. The point of this thread is not for me to point out something that I know that others do not, but more to explore 'aloud'/in public what we know about these effects.

From what I have now read about SDT and the memories it has jogged of the courses I did study, it appears to help explain the science behind overt decision making, an ABX perhaps, rather than an understanding of what is happening before we have fully processed our perception of signal and noise to make a determination.

Go to the top of the page
+Quote Post
Kees de Visser
post Apr 2 2012, 11:40
Post #81





Group: Members
Posts: 648
Joined: 22-May 05
From: France
Member No.: 22220



QUOTE (icstm @ Mar 30 2012, 11:25) *
I did not mean actually using a giant probe, I meant using a sensor to test earlier in the chain.
There are methods that don't hurt, but AFAIK they are mostly used to diagnose hearing loss and are probably not accurate enough for our purpose.
http://en.wikipedia.org/wiki/Auditory_brainstem_response
QUOTE
The auditory brainstem response (ABR) is an auditory evoked potential extracted from ongoing electrical activity in the brain and recorded via electrodes placed on the scalp.
...
The ABR is used for newborn hearing screening, auditory threshold estimation, intraoperative monitoring, determining hearing loss type and degree, and auditory nerve and brainstem lesion detection.
Go to the top of the page
+Quote Post
greynol
post Apr 2 2012, 17:29
Post #82





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



QUOTE (icstm @ Apr 2 2012, 02:23) *
From what I have now read about SDT and the memories it has jogged of the courses I did study, it appears to help explain the science behind overt decision making, an ABX perhaps, rather than an understanding of what is happening before we have fully processed our perception of signal and noise to make a determination.

The point (at least in part) is that one needs to be aware of whether the portion of the signal in question even makes it far enough along to be processed by whatever level of conscious state.

This post has been edited by greynol: Apr 2 2012, 17:49


--------------------
Placebophiles: put up or shut up!
Go to the top of the page
+Quote Post
Woodinville
post Apr 3 2012, 09:20
Post #83





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (greynol @ Apr 2 2012, 09:29) *
QUOTE (icstm @ Apr 2 2012, 02:23) *
From what I have now read about SDT and the memories it has jogged of the courses I did study, it appears to help explain the science behind overt decision making, an ABX perhaps, rather than an understanding of what is happening before we have fully processed our perception of signal and noise to make a determination.

The point (at least in part) is that one needs to be aware of whether the portion of the signal in question even makes it far enough along to be processed by whatever level of conscious state.


Indeed, and then it is possible to move on to level-roving experiments, and see how a layer of memory and its time-extent can be defined, then move on to more complex features, and see how long their memory lasts, and so on.

This is, of course, not all discussed in one place, it's a summary of 1000's of experiments and lots and lots of reading, gathered over 40 years of working in the field. I realize that's hard to convey in a few paragraphs. sad.gif

I was thinking about writing a book, but I kind of like being retired, too...


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
icstm
post Apr 3 2012, 10:44
Post #84





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



QUOTE (greynol @ Apr 2 2012, 16:29) *
The point (at least in part) is that one needs to be aware of whether the portion of the signal in question even makes it far enough along to be processed by whatever level of conscious state.
bingo! So if we have a good knowledge of this, then that provides excellent information.

Maybe I approached this thread incorrectly, but my starting point is that from what I had read here and on many other forums is that no one had created a cogent argument that put to bed issues that I would have thought should not be debated.

The fact that there are still "golden ears" posts suggests that there is room for debate. Now where I think I have been clear is to say that I am not trying to give weight to their assurtions, but I have been trying to create a more compelling structure to refute their claims, or hold up a hand a say, though I do not beleive in your claim and though I think it is a flying spaghetti monster, I do not have irrefutable proof to show you.
Go to the top of the page
+Quote Post
greynol
post Apr 3 2012, 15:27
Post #85





Group: Super Moderator
Posts: 10000
Joined: 1-April 04
From: San Francisco
Member No.: 13167



I wouldn't necessarily assume that discussions remain open because there is still ground for debate that is not just faith-based.

This post has been edited by greynol: Apr 3 2012, 15:28


--------------------
Placebophiles: put up or shut up!
Go to the top of the page
+Quote Post
Woodinville
post Apr 3 2012, 23:32
Post #86





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (icstm @ Apr 3 2012, 02:44) *
The fact that there are still "golden ears" posts suggests that there is room for debate.


Another suggestion would be to re-read the discussion at audioskeptic.blogspot.com (yes, that's mine) and especially concentrate on the last part. This should explain why we will always have golden ears and the like, I dare say, quite clearly.


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
Paulhoff
post Apr 4 2012, 00:35
Post #87





Group: Members
Posts: 106
Joined: 3-June 05
From: Coconut Creek Fl
Member No.: 22486



QUOTE (Woodinville @ Apr 3 2012, 04:20) *
I was thinking about writing a book, but I kind of like being retired, too...


An audio book for the Layman without all the WOO-WOO.

How it all really works.

That would be good.

Paul

smile.gif smile.gif smile.gif


--------------------
"Reality is merely an illusion, albeit a very persistent one." Albert Einstein
Go to the top of the page
+Quote Post
icstm
post Apr 4 2012, 10:24
Post #88





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



@ 'Woodinville'
Actually I am completely aware of part 4 and is something I completely agree with. However that does not address the gap in my knowledge that means I do not feel I have that cogent argument I previously referred to.

Dare I say you have not sure gap given your years of experience. I have only recently become interested in this again ten years after I was looking at anything like this during university.

However given that there is no postive tests from ABX testing and given comparisons of waveforms so that other parts of the audio/processing system are likely to impact the sources to a greater extent than the delta in sources themselves, I am reasonably happy (though with still a nagging feeling) that no one can construct an argument that they can show that there is a difference.

That nagging feeling is that given the potential set of tools and procedures at our disposal, I would have hoped we can show more that what has at least been discussed in this thread.
Go to the top of the page
+Quote Post
Woodinville
post Apr 4 2012, 12:03
Post #89





Group: Members
Posts: 1402
Joined: 9-January 05
From: JJ's office.
Member No.: 18957



QUOTE (icstm @ Apr 4 2012, 02:24) *
However given that there is no postive tests from ABX testing and given comparisons of waveforms so that other parts of the audio/processing system are likely to impact the sources to a greater extent than the delta in sources themselves, I am reasonably happy (though with still a nagging feeling) that no one can construct an argument that they can show that there is a difference.


Huh? Could you rephrase that?

Under good conditions, DBT's can resolve sounds down to within ->||<- of the actual noise level of the atmosphere at your ear drum. Signal detection works a bit better for those, of course, although it is also a type of blind testing.

As to 'other parts' I have no idea what you're thinking of. An ABX test uses the same equipment for all tests, EXCEPT for the specific part under test. So there aren't going to be variations in the other parts of the system, unless they are caused by the part under test.

What does comparisons of waveforms have to do with anything? SNR is mostly useless, after all. (Well, if you have 110dB SNR you're probably good, and under 6dB is bad, but between at least 6dB and 70-80dB you can easily make two signals one of which you can detect the error in, the other one which you can't.)

So, could you rephrase this in a a fashion a native english speaker can digest?


--------------------
-----
J. D. (jj) Johnston
Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Apr 4 2012, 13:01
Post #90





Group: Members
Posts: 3647
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (icstm @ Apr 4 2012, 05:24) *
However given that there is no postive tests from ABX testing and given comparisons of waveforms so that other parts of the audio/processing system are likely to impact the sources to a greater extent than the delta in sources themselves, I am reasonably happy (though with still a nagging feeling) that no one can construct an argument that they can show that there is a difference.


What is this "...given that there is no postive tests from ABX testing..." stuff all about?

There are tons of positive results from ABX tests.

The general rule for obtaining positive results from ABX tests is to select your tests from the realm of things that we know by other means to be audible. If you use ABX to double check a good recent textbook's statements about the thresholds of audibility for yourself, you might do a little better or a little worse, but that's the nature of small samples and particularly samples of 1. They vary a bit.

Let's look at the profile of your typical ABX test that the High End audio publications would and has complained about. In general, they don't provide any information about the technical performance of the equipment being compared, or if they do, a person who is well-informed about the human thresholds of hearing will probably dismiss the test on the grounds that it would require that the thresholds of hearing be orders of magnitude smaller then we know that they are.

In short, the ABX critics usually fault ABX just as sensibly as we might fault a champion runner for not being able to beat the current world's record by several orders of magnitude. You can't run the mile in 2 seconds? What sort of Olympic runner are you? Obviously sir, you are a fraud and completely undeserving of your honors. ;-)

Go to the top of the page
+Quote Post
Paulhoff
post Apr 4 2012, 16:36
Post #91





Group: Members
Posts: 106
Joined: 3-June 05
From: Coconut Creek Fl
Member No.: 22486



QUOTE (Arnold B. Krueger @ Apr 4 2012, 08:01) *
In short, the ABX critics usually fault ABX just as sensibly as we might fault a champion runner for not being able to beat the current world's record by several orders of magnitude. You can't run the mile in 2 seconds? What sort of Olympic runner are you? Obviously sir, you are a fraud and completely undeserving of your honors. ;-)


Also, we know we can hear (something of your choice) but with ABX testing we don't, so it is the testing and not us that is wrong, therefore ABX testing fails.

Paul

smile.gif smile.gif smile.gif


--------------------
"Reality is merely an illusion, albeit a very persistent one." Albert Einstein
Go to the top of the page
+Quote Post
icstm
post Apr 5 2012, 11:32
Post #92





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



QUOTE (Woodinville @ Apr 4 2012, 11:03) *
Under good conditions, DBT's can resolve sounds down to within ->||<- of the actual noise level of the atmosphere at your ear drum. Signal detection works a bit better for those, of course,

... So, could you rephrase this in a a fashion a native english speaker can digest?


Not sure about what you last line means as I am from the UK and many here seem to be from the States smile.gif

But I am interested in your statement "... at your ear drum". I have clearly mis-understood what ABX is testing in that case, as I thought the comparison is being determined within the cortex, even if it is using stimuli from the ear drum and much processing between.
Go to the top of the page
+Quote Post
2Bdecided
post Apr 5 2012, 11:47
Post #93


ReplayGain developer


Group: Developer
Posts: 5059
Joined: 5-November 01
From: Yorkshire, UK
Member No.: 409



QUOTE (icstm @ Apr 5 2012, 11:32) *
But I am interested in your statement "... at your ear drum". I have clearly mis-understood what ABX is testing in that case, as I thought the comparison is being determined within the cortex, even if it is using stimuli from the ear drum and much processing between.
He's telling you that, for certain tasks, that processing isn't doing much worse than a perfect microphone (at the ear drum) + perfect ADC + perfect computer could.

He's also trying to explain that, for many tasks, we know pretty well what's happening in the mechanical and transducer parts of the ear - and the results you get from asking people what they can and cannot hear match predictions of what those transducers (hair cells) in the ear pick up very well. So well that, for many tasks, that "much processing between" can just be considered as a near-optimal computer processing the signals from the hair cells.

In short, almost all the losses are in the mechanical and transducer part of the ear. "The rest" is acting like a very good computer, trying to extract something meaningful from those signals. It doesn't lose much. In fact, it's amazing how much it does pick out of the rotten signals it gets from the ear.


So you've got experiments from asking people what they hear, simulations of what's in the ear, experiments from probing real ears, known levels of atmospheric and in-ear noise, a century of scientific data, etc etc etc - all these agree. Then you've got audiophiles claiming that things which they shouldn't be able to hear (and that multiple failed ABX tests suggest they can't hear) are easily and obviously audible.

If you believe the latter camp, I've got some great magic sticks that I'd like to sell you. They just look like sticks, but they'll bring you health, wealth and happiness, not to mention marital bliss. A bargain at £100 a pair.

Cheers,
David.

This post has been edited by 2Bdecided: Apr 5 2012, 11:49
Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Apr 6 2012, 16:32
Post #94





Group: Members
Posts: 3647
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (Paulhoff @ Apr 4 2012, 11:36) *
QUOTE (Arnold B. Krueger @ Apr 4 2012, 08:01) *
In short, the ABX critics usually fault ABX just as sensibly as we might fault a champion runner for not being able to beat the current world's record by several orders of magnitude. You can't run the mile in 2 seconds? What sort of Olympic runner are you? Obviously sir, you are a fraud and completely undeserving of your honors. ;-)


Also, we know we can hear (something of your choice) but with ABX testing we don't, so it is the testing and not us that is wrong, therefore ABX testing fails.

Paul

smile.gif smile.gif smile.gif



;-)

Right, which reminds me of the discussion of DBTs that was ongoing at the time we invented ABX for audio. People like Shanefield were using the word Epistemology, which is the study of how we know what we believe.

Do we believe certain things because that is what we think, or do we believe those things because of some formal, insightful methodology that we use to determine such things?

I sometimes call what the GEs do "Naive Perception". People who have any understanding of illusions, understand that perceiving is not necessarily believing. In the study of perception, the difference between an illusion and a reliable perception is just that, reliability. The illusory lake in the desert or wet spots on hot pavement disappear when we change our viewing point. The real lake about 2 miles from me right now does not disappear when I look at it from any reasonable viewing point. If I jump into it I get wet. None of this is rocket science to anybody but small children.

Quote from Dan Shanefield's CV

"
Created a "meme," in other words, a newly-created idea that continues to spread throughout the world on its own. This meme is the "equalized double-blind" listening comparison of audio components, which is now used worldwide. (See D. J. Shanefield, High Fidelity, March 1980, page 57. Also published in Boston Audio Society Speaker, Nov. 1974 and June 1975. Listed as the earliest double-blind audio articles: see Refs. 27 & 28 in S. Lipshitz et al., J. Audio Engrng. Soc., Vol. 29, July 1981.)

Introduced the graphic equalizer to the general public, with first article on this component in million-reader magazines (see for example cover story by D. J. Shanefield in Stereo Review, May 1976, referred to again in Stereo Review, May 1996, page 112).
"
Go to the top of the page
+Quote Post
Arnold B. Kruege...
post Apr 6 2012, 16:44
Post #95





Group: Members
Posts: 3647
Joined: 29-October 08
From: USA, 48236
Member No.: 61311



QUOTE (icstm @ Apr 5 2012, 06:32) *
QUOTE (Woodinville @ Apr 4 2012, 11:03) *
Under good conditions, DBT's can resolve sounds down to within ->||<- of the actual noise level of the atmosphere at your ear drum. Signal detection works a bit better for those, of course,

... So, could you rephrase this in a a fashion a native english speaker can digest?


But I am interested in your statement "... at your ear drum". I have clearly mis-understood what ABX is testing in that case, as I thought the comparison is being determined within the cortex, even if it is using stimuli from the ear drum and much processing between.


What sets the ultimate threshold of hearing, even if we mistakenly presume that the ear has infinite sensitivity?

The ultimate threshold of hearing would be determined by the that is noise generated by the thermal agitation of the molecules in the atmosphere. The little buggers randomly bounce off our eardrums, right? They must be causing a steady rush of noise in our ears that masks other potentially audible sounds that are appreciably softer. That random noise due to the motion of the air molecules would limit our ability to hear softer sounds, even if our ears themselves had unlimited sensitivity, right?

Several relevant factoids.

One is that few if any of us have any perception of the noise created by the molecules atmosphere bouncing off of our ear drums, even if we are in a very, very quiet room.

Another is what JJ pointed out above, which is that we can do DBTs where people successfully reliably detect noises that are just a tad louder than the noise created by the atmosphere molecules bouncing off of our ear drums.

Note that the noise level in even the quietest rooms we frequent are 20-40 dB louder than the noise created by the atmosphere molecules bouncing off of our ear drums.

Go to the top of the page
+Quote Post
icstm
post Apr 10 2012, 11:19
Post #96





Group: Members
Posts: 121
Joined: 25-January 12
Member No.: 96698



Hope everyone had a good Easter break.
Reading those last couple of posts I found very useful.
@David - forming the argument the way you did should surly put the other camp to bed... shame it does not.
Still it does mean there is money to be made from your magic sticks, I just hope you have a stomach that can handle selling them.
Go to the top of the page
+Quote Post

4 Pages V  « < 2 3 4
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 24th July 2014 - 17:57