Conclusive "Proof" that higher resolution audio sounds different

[jkeny]

There's no need to retract what I haven't said, John. Odd to quote myself, but here goes:

I didn't say that you broadly denied the validity of ABX testing, John.

Ok, if that's your retraction, then so be it?

And I didn't realize that you were applying those controls only in the event of a negative result, because I honestly didn't expect you to take such an absurd position. Controls are applied before testing, to all tests, before you have a result. Applying controls after the result, to a specific result, is the opposite of control, it is influence. I know it's a hard pill to swallow but it's science & sometimes science is tough.

Tim

Ah, Tim, please follow the logic - you keep making these snide swipes at me "absurd position" & trying to paint my position as somehow ridiculous - debate the topic without the use of emotive language, please.

As said to you already, a positive ABX test proves the test (because of the statistical analysis - it's the trump card) - despite whatever controls were not in place! People then want to put a spotlight on the lack of controls after the test results who were quite happy to accept null results before without question - witness IMD questions, resampler questions, timing slew questions. The fact that a positive result is returned even though some controls were not in place is also no reason to jump to the conclusion that the controls are therefore not needed - it's not logical.

A negative result throws an immediate spotlight on the controls in place for the test because there is no statistical trump card. A negative result could be returned because of genuine chance, because the equipment was not resolving enough, because the participants were not adept at hearing the differences, because the test material was not suitable, because the test was a pressured environment, because......... There are many reason & no trump card to play!! Deal with this before the test if you want it to be considered worthy or deal with the consequences of not looking after controls, it's up to the organisers of the test. Don't say that it's not a fair playing pitch or it's absurd - science is a bitch!

This previous unquestioning acceptance of null results is what you & others are now having to face up to & hopefully analyse calmly but so far this hasn't happened.

 

[Orb]

First let me make sure I understand what you're saying here -- "Unless you doubt on Amir report of the facts." That seems to be asking if I doubt Amir's report of his positive ID. I do not. Read the post above that you quoted and responded to and it will make that clear.

Subtle? I would call anything that requires someone to know what differences to listen for, and to listen to specific short passages that highlight those differences in order to hear them is subtle. But your definition may vary.

Tim

Umm,
I think like 50 pages ago many of us pointed out one cannot make such assumptions.
The ONLY conclusion one can take from the listening test is that some did detect differences, we do not the reason why and we DO NOT know implications this has for consumer listening behaviours.
Your conclusion sort of breaks down in this example; illusionists (such as Derrin Brown) that use suggestion to influence the person who is aware and watching-listening but does not cognitively recognise it/programming happening; cases include getting them think of a number/draw a picture/etc where they would go out of their way to think of doing something they would not expect the illusionist to match.

These type of advanced suggestion are very complex and subtle, even watching back recording multiple times only the simple examples can be identified by the non-trained.
The point being with music we may be influenced in similar ways and what seems very small (but shown to be audibly detectable) in absolute terms of perception may be more from a cognitive-behaviour focus.
Emphasis on "may" because there is still a lot of research relating to human behaviour and perception-music-maths-language-emotion relationship.

Unfortunately the case you are putting forward would need a specific listening behaviour study to get anywhere close to a fact for your context, I say unfortunately because these would be freaking complex tests-environments (need to capture and weigh biases-influences-etc), which is why I guess most modern studies look to use tests and ways to measure via brain activity/connections.

The illusionist example is not perfect, just an extreme to help, but the point is this ABX (context this thread) only proves there is a detectable difference between hirez-CD quality for some reason that may have implications with how digital audio is handled and may or may not influence listening behaviour; we know now trained listeners (include informal via forums on what to listen for and methodology) can identify something happening that in the past has been presented as being impossible and nothing more beyond this can be taken from the test IMO.
Same way tolerances-thresholds need to be correlated to behaviour patterns, especially when the source-content activates emotional-satisfaction levels within us.

Thanks
Orb

 

[amirm]

That's not stress or some other trivial factor invalidating the test, Amir, that's a difference so subtle that only a listener who knows what to listen for and where to listen can hear it. To all others, the files are transparent. The fact that you, and a handful of people who were told what to listen for, did hear a difference means that the test was, indeed, effective. The fact that the files were transparent to those who were not trained to ID the differences, means that the differences were extremely subtle. It sounds like semantics, but it is perception.

Tim

That is not what I meant Tim. I said that as differences get smaller, but still audible, it becomes easier and easier to construct blind tests where no difference is heard. Even by those who could otherwise hear it. Yes, there is a point at which only trained listeners may be able to hear the difference but I am talking about stage before that.

 

[RogerD]

That is not what I meant Tim. I said that as differences get smaller, but still audible, it becomes easier and easier to construct blind tests where no difference is heard. Even by those who could otherwise hear it. Yes, there is a point at which only trained listeners may be able to hear the difference but I am talking about stage before that.

So Amir, In scientific jargon what was the difference attributable to or what exactly did the spectrum analysis reveal between the two files. I'm not interested in the audible test proceduce,just the spectrum analysis.

 

[Phelonious Ponk]

Umm,
I think like 50 pages ago many of us pointed out one cannot make such assumptions.
The ONLY conclusion one can take from the listening test is that some did detect differences, we do not the reason why and we DO NOT know implications this has for consumer listening behaviours.
Your conclusion sort of breaks down in this example; illusionists (such as Derrin Brown) that use suggestion to influence the person who is aware and watching-listening but does not cognitively recognise it/programming happening; cases include getting them think of a number/draw a picture/etc where they would go out of their way to think of doing something they would not expect the illusionist to match.

These type of advanced suggestion are very complex and subtle, even watching back recording multiple times only the simple examples can be identified by the non-trained.
The point being with music we may be influenced in similar ways and what seems very small (but shown to be audibly detectable) in absolute terms of perception may be more from a cognitive-behaviour focus.
Emphasis on "may" because there is still a lot of research relating to human behaviour and perception-music-maths-language-emotion relationship.

Unfortunately the case you are putting forward would need a specific listening behaviour study to get anywhere close to a fact for your context, I say unfortunately because these would be freaking complex tests-environments (need to capture and weigh biases-influences-etc), which is why I guess most modern studies look to use tests and ways to measure via brain activity/connections.

The illusionist example is not perfect, just an extreme to help, but the point is this ABX (context this thread) only proves there is a detectable difference between hirez-CD quality for some reason that may have implications with how digital audio is handled and may or may not influence listening behaviour; we know now trained listeners (include informal via forums on what to listen for and methodology) can identify something happening that in the past has been presented as being impossible and nothing more beyond this can be taken from the test IMO.
Same way tolerances-thresholds need to be correlated to behaviour patterns, especially when the source-content activates emotional-satisfaction levels within us.

Thanks
Orb

Orb? What are you talking about? That I don't doubt that Amir heard what he reports hearing, or that I would call anything requiring training and a very specific methodology (as this did in this example) to identify subtle? Notice the "I" in both of those. Either way, I don't recall any of you telling me, 50 pages or so ago, that I could neither presume to trust Amir's judgement nor have an opinion of what is subtle.

tim

 

[Phelonious Ponk]

Ok, if that's your retraction, then so be it?
Ah, Tim, please follow the logic - you keep making these snide swipes at me "absurd position" & trying to paint my position as somehow ridiculous - debate the topic without the use of emotive language, please.

As said to you already, a positive ABX test proves the test (because of the statistical analysis - it's the trump card) - despite whatever controls were not in place! People then want to put a spotlight on the lack of controls after the test results who were quite happy to accept null results before without question - witness IMD questions, resampler questions, timing slew questions. The fact that a positive result is returned even though some controls were not in place is also no reason to jump to the conclusion that the controls are therefore not needed - it's not logical.

A negative result throws an immediate spotlight on the controls in place for the test because there is no statistical trump card. A negative result could be returned because of genuine chance, because the equipment was not resolving enough, because the participants were not adept at hearing the differences, because the test material was not suitable, because the test was a pressured environment, because......... There are many reason & no trump card to play!! Deal with this before the test if you want it to be considered worthy or deal with the consequences of not looking after controls, it's up to the organisers of the test. Don't say that it's not a fair playing pitch or it's absurd - science is a bitch!

This previous unquestioning acceptance of null results is what you & others are now having to face up to & hopefully analyse calmly but so far this hasn't happened.

A single positive proves, and a single negative immediately throws the methodology of the test into question...yeah, that science stuff is hard, John.

Tim

 

[jkeny]

A single positive proves, and a single negative immediately throws the methodology of the test into question...yeah, that science stuff is hard, John.

Tim

Yep, Tim, takes getting used to it - it's unforgiving.
But you are proving your complete lack of objectivity in this by your posts
It isn't a single positive result being reported from Arnyk's test
It wasn't a single negative result reported up to now
But even so, yes a single positive is statistically a very strong result in favour of there being a difference between samples
After this positive result thousands of previous negative results is a strong indicator that proper controls were not being used

Sorry, Tim but you are kicking & screaming against something which is well known. Instead of all this arguing why not do what Max does?

 

[microstrip]

First let me make sure I understand what you're saying here -- "Unless you doubt on Amir report of the facts." That seems to be asking if I doubt Amir's report of his positive ID. I do not. Read the post above that you quoted and responded to and it will make that clear.

Subtle? I would call anything that requires someone to know what differences to listen for, and to listen to specific short passages that highlight those differences in order to hear them is subtle. But your definition may vary.

Tim

Tim,

Sorry it is not possible to debate with someone who systematically ignores the global sense of sentences or documents, and picks only small parts of them to manipulate the arguments. The sense of my post was clear, and you quoted in bold part of it , ignoring the main point to create a fait divers. I tried to use what you seem to ask - science taken from ITU recommendations and know facts about ABX tests. Sorry I do not have the time to explain very basic level statistical arguments of ABX in detail - there are many places documenting them.

 

[maxflinn]

Thousands of null results are a very strong indication that the differences people claim to hear sighted are not real. I can't ever recall a blind-test where the participants reported no differences following the sighted listening that always precedes the blind-testing at these events.

The removal of knowledge is what causes the blind nulll results. This is obvious.

Null results are what they are and do not indicate a lack of controls. All in my humble, logical opinion, of course.

 

[microstrip]

Micro, do you consider a null result to be due to a poorly carried out test when the participants reported differences sighted first?

Also, have you any evidence to support your view that null results are usually the result of poorly carried out tests?

Thanks.

Max,

There seems to be some confusion between the sense of null or negative and not valid in your question. I can not see any systematic connection between nulls and sighted tests.

A null is never 100% valid - but if you have controls its comprehensiveness will be much wider.

The asked evidence is given by the non existence of tests resulting in any significant positive identifications - and in part believing that the people who wrote specifications asking for controls know what they are addressing. Unfortunately these tests we refer are never properly documented, and any post asking for details is never answered. The only rigorous challenge about "small differences" (electronics, cables and sources) I know about was the Quad challenge, properly reported and analyzed in Electronics and Wireless World many decades ago.

 

[maxflinn]

Micro, thanks for the reply. I'll respond tomorrow when at my PC. Using a phone at the moment and posting is awkward.

 

[Phelonious Ponk]

Yep, Tim, takes getting used to it - it's unforgiving.
But you are proving your complete lack of objectivity in this by your posts
It isn't a single positive result being reported from Arnyk's test
It wasn't a single negative result reported up to know
But even so, yes a single positive is statistically a very strong result in favour of there being a difference between samples
After this positive result thousands of previous negative results is a strong indicator that proper controls were not being used

Sorry, Tim but you are kicking & screaming against something which is well known. Instead of all this arguing why not do what Max does?

John, you should have just explained this in this simple, straightforward manner a long time ago. Then we would have understood all along that no matter what happens, you're right, or the result is wrong. That would have made it so much simpler.

Tim

 

[jkeny]

John, you should have just explained this in this simple, straightforward manner a long time ago. Then we would have understood all along that no matter what happens, you're right, or the result is wrong. That would have made it so much simpler.

Tim

As, Micro said - it looks like the basics escape you, Tim, I'm afraid

 

[Phelonious Ponk]

As, Micro said - it looks like the basics escape you, Tim, I'm afraid

Pretty much everybody but you and micro, I'm afraid.

Tim

 

[still-one]

Over 800 posts in this thread and yet I have not read where one individual has changed his position based on these posts. I sure as heck am not going back to double check.

 

[microstrip]

Over 800 posts in this thread and yet I have not read where one individual has changed his position based on these posts. I sure as heck am not going back to double check.

And you? Do you think it was proved or not?
Perhaps it is time for a poll!

 

[thedudeabides]

Over 800 posts in this thread and yet I have not read where one individual has changed his position based on these posts. I sure as heck am not going back to double check.

The dog continues to chase it's tail.

WOOF!

 

[esldude]

And you? Do you think it was proved or not?
Perhaps it is time for a poll!

We need no polls when you can grab the files and find out for yourself.

I was able to abx the original Jangling keys file. But the resampling was less than good. Resampled with a good software I could no longer detect a difference. The original AVS files were different in level by .2db known to be audible in blind tests. Nothing proved there as it was something already known.

Here is the FFT with original resampling of the Jangling keys with the original 96 khz subtracted from the resampled 44.1 version.


Here is the FFT of the difference with better resampling.



The difference file with the Jangling keys and resampled version had faint audible music. When resampled with better software there is nothing other than very low level noise. Even amplified there is noise with no trace of music at all.

Similar results for the AVS files. I could not ABX them once they were resampled with a quality resampler. And the residual had no audible sound other than noise that needed heavy amplification to be heard at all.

There were ultrasonic sounds left of course. One could play back the file at a different sample rate shifting the ultrasonic range down below 15 khz and hear it. It was low in level, but I can only surmise I can't hear above 20 khz (big surprise).

Amirm also resampled and still ABX'd them, but I don't know which resampler he used nor what sort of FFT or difference file he got from his procedures.

Further I would still think the most likely thing audible here is dither from resampling and changing from 24 to 16 bit. Would be interesting if Amirm would resample to 44/24 and see if he can ABX that successfully. If you change sample rates with both at 24 bit the difference of such an operation is at extremely low levels. The result is like this with residual difference below 20 khz in the -175-180 db range.

 

[RogerD]

We need no polls when you can grab the files and find out for yourself.

I was able to abx the original Jangling keys file. But the resampling was less than good. Resampled with a good software I could no longer detect a difference. The original AVS files were different in level by .2db known to be audible in blind tests. Nothing proved there as it was something already known.

Here is the FFT with original resampling of the Jangling keys with the original 96 khz subtracted from the resampled 44.1 version.
View attachment 16825

Here is the FFT of the difference with better resampling.

View attachment 16826

The difference file with the Jangling keys and resampled version had faint audible music. When resampled with better software there is nothing other than very low level noise. Even amplified there is noise with no trace of music at all.

Similar results for the AVS files. I could not ABX them once they were resampled with a quality resampler. And the residual had no audible sound other than noise that needed heavy amplification to be heard at all.

There were ultrasonic sounds left of course. One could play back the file at a different sample rate shifting the ultrasonic range down below 15 khz and hear it. It was low in level, but I can only surmise I can't hear above 20 khz (big surprise).

Amirm also resampled and still ABX'd them, but I don't know which resampler he used nor what sort of FFT or difference file he got from his procedures.

Further I would still think the most likely thing audible here is dither from resampling and changing from 24 to 16 bit. Would be interesting if Amirm would resample to 44/24 and see if he can ABX that successfully. If you change sample rates with both at 24 bit the difference of such an operation is at extremely low levels. The result is like this with residual difference below 20 khz in the -175-180 db range.
View attachment 16827

Do these measure harmonic structure. Noise has a signature that is high frequency smearing or harmonic distortion,it effects clarity.

 

[esldude]

Do these measure harmonic structure. Noise has a signature that is high frequency smearing or harmonic distortion,it effects clarity.

Yes it would. In this particular case the FFT is showing bins approximately 11 hz wide. Each section of 11 hz makes up the graph. 1 khz would show a spike near 1khz and another at 2khz or 3 khz if there were harmonics.

Noise need not be high frequency, it can have any frequency. When I say noise is audible here in the difference, it is only audible with amplification of some 40 decibels when good resampling was used. If you just listen to the file on its own, you hear silence. With amplification you hear something much like interstation hiss from FM with quieting turned off.