Debate Thread: Scott's Hi-res Audio Test - Page 73 - AVS Forum
Forum Jump: 
 
Thread Tools
Old 07-20-2014, 04:11 PM
Senior Member
 
koturban's Avatar
 
Join Date: Jan 2007
Posts: 385
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 155 Post(s)
Liked: 102
Quote:
Originally Posted by amirm View Post
Someone linked to remarkably similar results to the experiment here on WBF Forum. Here is the original post:

http://www.hydrogenaud.io/forums/ind...t=#entry713014

----------
foo_abx 1.3.4 report
foobar2000 v1.0.3
2010/06/07 18:25:51

File A: C:\Dokumente und Einstellungen\Christian\Desktop\16-441to24-96.wav
File B: C:\Dokumente und Einstellungen\Christian\Desktop\24-96.wav

18:25:51 : Test started.
18:26:44 : 01/01 50.0%
18:27:09 : 02/02 25.0%
18:27:26 : 03/03 12.5%
18:27:44 : 04/04 6.3%
18:28:00 : 05/05 3.1%
18:28:21 : 06/06 1.6%
18:28:38 : 07/07 0.8%
18:28:52 : 08/08 0.4%
18:29:06 : 09/09 0.2%
18:29:17 : 10/10 0.1%
18:29:33 : 11/11 0.0%
18:29:46 : 12/12 0.0%
18:30:03 : 13/13 0.0%
18:30:26 : 14/14 0.0%
18:30:42 : 15/15 0.0%
18:31:04 : 16/16 0.0%
18:31:08 : Test finished.

----------
Total: 16/16 (0.0%)


I was so surprised to see Arny in that thread and hence having knowledge of other positive results of the nature we have had in this thread..

Looks like his testing of 96/24 against 44/16 were done using speakers:"My hearing tops at about 17 kHz. Still I can hear pretty clear difference over a pair of Elac FS 607 X-Jet speakers, which are rated for 28-50000Hz (IEC 268-5)"
Did you perform the IM distortion test?
koturban is offline  
Sponsored Links
Advertisement
 
Old 07-20-2014, 04:28 PM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by koturban View Post
Did you perform the IM distortion test?
No one has outlined any such test. Arny is cooking up something:

Quote:
Originally Posted by arnyk View Post
I've been investigating this problem and am working on a listening test that is a simple, unmistakable go/nogo test for monitoring systems. It has surprised me, because while I expected that nonlinearity in the electronics would be the most likely problem, it appears that transducer problems can also be very strong.
But he is not finished and one has to wonder how he can create a test for an artifact that he himself can't hear! But that is why he gets paid the big bucks. He can do the impossible .

Did you perform any listening tests? Those have been ready and available for quite some time now.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-20-2014, 04:40 PM
Senior Member
 
koturban's Avatar
 
Join Date: Jan 2007
Posts: 385
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 155 Post(s)
Liked: 102
Quote:
Originally Posted by amirm View Post
No one has outlined any such test. Arny is cooking up something:



But he is not finished and one has to wonder how he can create a test for an artifact that he himself can't hear! But that is why he gets paid the big bucks. He can do the impossible .

Did you perform any listening tests? Those have been ready and available for quite some time now.
Did you read the thread you linked or are you just trolling for positive results from the HA forum to feed your confirmation bias?

The answer is there.
koturban is offline  
Old 07-20-2014, 05:48 PM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by koturban View Post
Did you read the thread you linked or are you just trolling for positive results from the HA forum to feed your confirmation bias?
Positive results are positive results. We have a responsibility to disclose such exculpatory evidence to people if we are aware of them as opposed to making it look like no one has ever passed such a test. If we had done that, it would have saved us a ton of work and countless posts in this thread alone.

We look like corrupt politicians in the way we hide such evidence and then defend our conduct in the manner you post.

Anyway, you did not indicate what IM test you were asking me to run so I assumed you meant the one Arny mentioned in this thread. If you meant what he said in HA thread, then you should have been specific. But no, I did not plan to run that test until Arny provided some concrete data that his hypothesis for Intermodulation distortion has legs. This can't be a process where someone just throws out guesses left and right and the rest of us run like lemmings run a bunch of tests to prove of disprove their hypothesis. We may have two digit IQ but we are not that "stoopid".

Just to close this chapter anyway, I ran this test that Arny proposed there: "another, potentially safer test signal would be white or pink noise high pass filtered at 22 KHz. If there is excess nonlinear distoriton in your system you'll hear noise in the normal audio band. "

I created a blank file at 96 Khz/24-bit and filled it with white noise. I then high-pass filtered it at 24 Khz. At 22 Khz even with maximum filter order, it would still preserve spectrum below 20 Khz. I then played this. I hear nothing using my Shure IEM. Not only that, our two dogs who can hear and recognize sounds thousand times better than any of us, would not react whatsoever from 2 foot distance. Playing the white noise prior to filtering made both of them jump.

So there, you have both dog and human evidence that Arny's hypothesis doesn't hold.

Quote:
Originally Posted by kat
The answer is there.
No. Busy work and deflection was there .

Since you ignored my equal request of you, I will go ahead and put on record that you are refusing to perform any listening tests. This will remain on your academic record for at least 10 years. If at the end of that period you still have not run and reported on the results of any listening tests you have performed, the penalty mark will remain there for 10 more years and process repeated. So ignore my requests for running the tests in this thread at your peril.
Marc Wielage and stereoeditor like this.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-20-2014, 06:26 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by onlyconnect View Post
Ah, that's good I have Audition too though I don't pretend to be expert on all the different resampling options.


That said, I got an odd result. I made my own downsample to 16/44 with all the options disabled, and then upsampled back to 24/96. I then inverted my downsample and mixed it with the original full band version. I get a perfect null for frequencies below c 20K.


However if I do the same with the supplied 16/44 sample (which I understand was used for the ABX) it doesn't null well at all. Maybe the two files are not perfectly synched?
You may not realize this, but a perfect null is an erroneous result.

At the very least, a word length reduction such as that from 24 bits to 16 must be dithered, and the addition of dither makes it impossible for the files to null below 20 KHz since that is the frequency range where the dither is applied.

In addition downsampling should be pre-filtered, then downsampled and post-filtered in order to avoid aliasing. The very minor timing and frequency response variations caused by the pre and post filtering again makes it impossible for the files to null perfectly.
arnyk is offline  
Old 07-20-2014, 06:32 PM
Senior Member
 
koturban's Avatar
 
Join Date: Jan 2007
Posts: 385
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 155 Post(s)
Liked: 102
Quote:
Originally Posted by amirm View Post
Positive results are positive results. We have a responsibility to disclose such exculpatory evidence to people if we are aware of them as opposed to making it look like no one has ever passed such a test. If we had done that, it would have saved us a ton of work and countless posts in this thread alone.
I replied to your post within its context. ie, what you posted. It's disappointing, but not the least surprising, that you continue to evade even the most straightforward questions.



Quote:
Originally Posted by amirm View Post
We look like corrupt politicians in the way we hide such evidence and then defend our conduct in the manner you post.
Who is this "we"?



Quote:
Originally Posted by amirm View Post
Anyway, you did not indicate what IM test you were asking me to run so I assumed you meant the one Arny mentioned in this thread. If you meant what he said in HA thread, then you should have been specific. But no, I did not plan to run that test until Arny provided some concrete data that his hypothesis for Intermodulation distortion has legs. This can't be a process where someone just throws out guesses left and right and the rest of us run like lemmings run a bunch of tests to prove of disprove their hypothesis. We may have two digit IQ but we are not that "stoopid".

Just to close this chapter anyway, I ran this test that Arny proposed there: "another, potentially safer test signal would be white or pink noise high pass filtered at 22 KHz. If there is excess nonlinear distoriton in your system you'll hear noise in the normal audio band. "

I created a blank file at 96 Khz/24-bit and filled it with white noise. I then high-pass filtered it at 24 Khz. At 22 Khz even with maximum filter order, it would still preserve spectrum below 20 Khz. I then played this. I hear nothing using my Shure IEM. Not only that, our two dogs who can hear and recognize sounds thousand times better than any of us, would not react whatsoever from 2 foot distance. Playing the white noise prior to filtering made both of them jump.



So there, you have both dog and human evidence that Arny's hypothesis doesn't hold.
Anecdote.


Quote:
Originally Posted by amirm View Post
No. Busy work and deflection was there .

Since you ignored my equal request of you, I will go ahead and put on record that you are refusing to perform any listening tests. This will remain on your academic record for at least 10 years. If at the end of that period you still have not run and reported on the results of any listening tests you have performed, the penalty mark will remain there for 10 more years and process repeated. So ignore my requests for running the tests in this thread at your peril.
That, and a dollar will buy a cup of coffee.

It's your reputation here, as the burden of proof has always been on you. I'm not the one shilling overpriced audio equipment.

My participation (or non-participation) has absolutely no bearing on your burden of proof.
Marc Wielage and Ratman like this.
koturban is offline  
Old 07-20-2014, 06:47 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by amirm View Post
No one has outlined any such test. Arny is cooking up something:
My starting point is a simple two tone IM test. The two tones need to be at frequencies for which audibility is very faint or non existent, such as in the 30-40 KHz+ region. The two tones should be different in frequency by an amount that results in a highly audible spurious response if there is any IM.

Our reference is the Fletcher Munson curves for audibility versus frequency:



The frequencies in the 1 KHz to 6 Khz range are very audible with audiblity peaking just below 4 KHz.

Therefore the tones should be from 1 KHz to 6 Khz apart with a 4 KHz difference being maximally audible.

This is a test of 30 and 31 Khz with 2% second order IM present:




Please notice the relatively huge spike at 1 KHz.

Therefore I have been experimenting with tones at 40 and 44 KHz for example.

Amir has been falsely claiming that I can't hear anything above 8 KHz, which is false and not indicated by anything that I've said but we all know about Amir's affinity for hyperbole. ;-)


It is well known and I have posted on AVS several times that it is very conceivable that one will not hear frequencies in a context where they are masked, even when they are clearly audible when they are not masked. Frequencies above say 8 KHz can easily be masked by frequencies below 8 KHz. Therefore their absence in context may not be heard even though pure tones at the same frequencies may be heard.

The interesting thing is that I have had no difficulty obtaining positive results. I figured that since this was just a $1 chip (Realtek® ALC887 8-Channel High Definition Audio) why should I be surprised?

However, I measured the two tone distortion of this PC's audio interface by recording it with my Microtrak @ 2496. All spurious responses were if memory serves (FFT on request) 95 dB or better down, which I couldn't possibly hear.

I am provisionally concluding that the IM is in my Audio Technica headphones. When examining them I was surprised to find that they were my wife's ATH M30s not my ATH M50s. I am continuing to study the situation.

Quote:
Originally Posted by amirm View Post
But he is not finished and one has to wonder how he can create a test for an artifact that he himself can't hear! But that is why he gets paid the big bucks. He can do the impossible .
As I have shown above things that are completely and easily possible may be unexpected by people who don't understand how nonlinear distortion (or masking) actually works.
Attached Thumbnails
Click image for larger version

Name:	IM 30 & 31 KHz.png
Views:	246
Size:	36.8 KB
ID:	172426  

Last edited by arnyk; 07-20-2014 at 07:03 PM.
arnyk is offline  
Old 07-20-2014, 07:06 PM
AVS Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,360
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 229
Quote:
Originally Posted by GGA View Post
That is very interesting. How does recording at -20dBFS improve things and what does it improve? Like you said I was always told to get near zero. I am primarily archiving my vinyl to digital. Thanks.
The danger of maxing out the peak level is if your monitoring system does not accurately show you *reconstructed* levels -- including 'intersample overs' (levels that exceed 0dBFS, therefore clipping).
krabapple is offline  
Old 07-20-2014, 07:10 PM
AVS Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,360
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 229
Quote:
Originally Posted by rufus13 View Post
If you can hear an effect on headphones, but not your loudspeaker/room combination, it doesn't mean that the headphones are unfairly revealing. It is just the opposite: your loudspeakers are unfairly concealing.
No, not necessarily. Your loudspeakers *might* be fine.

Quote:
Originally Posted by rufus13 View Post
It mean that you have not perfected your room.
Yes, quite possibly, though 'perfection' is neither necessary or even attainable, except *perhaps* for one very narrow listening position.
krabapple is offline  
Old 07-20-2014, 07:26 PM
AVS Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,360
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 229
Quote:
Originally Posted by stereoeditor View Post
Krabapple was involved in a thread on Hydrogen Audio about 5 years ago, where I was being strenuously criticized for recommending Stereophile readers to use lossless rather than lossy compression. My point was that while MP3 and AAC at the higher bit rates might be sufficiently transparent with some recordings to satisfy a majority of listeners, you don't know ahead of ripping a recording if that will be true or not. In addition, what might be transparent to a particular listener now might not be as his listening abilities develop, or if he had learned what to listen for with lossy compressed music.
Neither I nor anyone else on HA never claimed that mp3s at even the 'best' bitrate/codec combo are transparent, full stop (i.e., impossible to differentiate from souce, by anyone), regardless of what either you duffers remember. And of course, 5 years ago, and today, *anyone* with a computer and an internet connection can experiment with mp3 encoding and ABX testing to determine what will be 'true or not' for theuir ears.

I still hold that any demonstration (one might say demonization) of mp3 that involves playing the difference signal for an mp3 versus source, as an indicator of mp3 sound is a joke and a feat of misdirection at best, and a lie at worst. Similarly any demo that involves comparing spectra. Similarly any 'demo' comparing mp3 to source that does not specify the codec and the bitrate and is not performed blind is a joke and sleight at best, a lie at worst.

Similarly any damn thing Mikey Fremer writes about lossy audio.

Quote:
Originally Posted by stereoeditor View Post
The only safe strategy to guarantee transparency and still reduce storage space was therefore to rip to a lossless format like ALAC or FLAC.
It is, more importantly, the only safe strategy to ensure that you can transcode to a lossy format as desired, without adding unnecessary artifacts that can accrue from lossy-->lossy.

Quote:
Originally Posted by stereoeditor View Post
You would have I thought I was recommending people cut off a hand, so strong was the criticism on HA of what I felt to be sensible, practical advice. Especially as storage was now ridiculously cheap compared to how expensive it was when MP3 became popular and how much fatter download pipes had become.

Seems incontrovertible.

And that was the point I made on HA, to universal condemnation.

I suggest people here take *your* spin on anything that happens on HA with teaspoon of salt, and investigate it on their own.

Last edited by krabapple; 07-20-2014 at 07:40 PM.
krabapple is offline  
Old 07-20-2014, 08:02 PM
AVS Special Member
 
krabapple's Avatar
 
Join Date: Apr 2003
Location: in a state bordered by Kentucky and Maine
Posts: 5,360
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 177 Post(s)
Liked: 229
Quote:
Originally Posted by stereoeditor View Post
Actually, I was thinking of this earlier thread, http://www.hydrogenaud.io/forums/ind...howtopic=71245 ,
Ah, the thread where Fremer came into the fray at full boil and JJ proceeded to thoroughly school him on lossy audio and testing. Priceless.

Edit: ooh, I quite forgot that later in the thread a fellow from Fraunhofer schooled JA himself on mp3
Marc Wielage likes this.

Last edited by krabapple; 07-20-2014 at 08:11 PM.
krabapple is offline  
Old 07-20-2014, 08:03 PM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by arnyk View Post
Amir has been falsely claiming that I can't hear anything above 8 KHz, which is false and not indicated by anything that I've said but we all know about Amir's affinity for hyperbole. ;-)
My deepest apologies. My English skills are often as bad as my IQ . I do know how to use the forum though and went back to your post and it said this:

Quote:
Originally Posted by arnyk View Post
I now struggle to hear the effects of an 8 KHz brick wall filter at normal listening levels.
Doesn't that say that if I get rid of everything above 8 Khz, you have a hard time telling that I have done that?

Quote:
Originally Posted by arny
It is well known and I have posted on AVS several times that it is very conceivable that one will not hear frequencies in a context where they are masked, even when they are clearly audible when they are not masked. Frequencies above say 8 KHz can easily be masked by frequencies below 8 KHz. Therefore their absence in context may not be heard even though pure tones at the same frequencies may be heard.
Masking applies to everyone. I thought you were saying that you fail the tests in this thread including your own because you simply can't hear high frequencies above 8 Khz. If you are now saying that this is due to masking, how come masking didn't stop me and others from hearing the degradations but did that to you???

Quote:
Originally Posted by arny
The interesting thing is that I have had no difficulty obtaining positive results. I figured that since this was just a $1 chip (Realtek® ALC887 8-Channel High Definition Audio) why should I be surprised?
Hey, we are surprised .

Quote:
Originally Posted by arny
However, I measured the two tone distortion of this PC's audio interface by recording it with my Microtrak @ 2496. All spurious responses were if memory serves (FFT on request) 95 dB or better down, which I couldn't possibly hear.
You can hear distortion down 95 db? Maybe we should resurrect those HDMI jitter discussions .

Quote:
Originally Posted by arny
As I have shown above things that are completely and easily possible may be unexpected by people who don't understand how nonlinear distortion (or masking) actually works.
Count me in that camp of being stupid about audio anytime you like Arny. That is why I am here. To learn from you what non-linear distortion or masking are.

What do you think of this as a teaching aid for masking?


Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-20-2014, 09:01 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by krabapple View Post
Ah, the thread where Fremer came into the fray at full boil and JJ proceeded to thoroughly school him on lossy audio and testing. Priceless.

Edit: ooh, I quite forgot that later in the thread a fellow from Fraunhofer schooled JA himself on mp3
I had tried to perform similar schooling earlier but JA was so accustomed to dissing me by then that he couldn't benefit. Seriously. The context (if memory serves) is that I saw an early draft of JAs article that involved algebraic subtraction of decoded MP3s from the .wav file used to create them, and pointed out the same problems that everybody else who is knowledgeable about perceptual coding would point out.

It was interesting to look at those old threads. I saw where I couldn't convince one of the HA regulars that aliasing is nonlinear distortion. Same problem as with JA.
arnyk is offline  
Old 07-20-2014, 09:22 PM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by arny

It is well known and I have posted on AVS several times that it is very conceivable that one will not hear frequencies in a context where they are masked, even when they are clearly audible when they are not masked. Frequencies above say 8 KHz can easily be masked by frequencies below 8 KHz. Therefore their absence in context may not be heard even though pure tones at the same frequencies may be heard.
Quote:
Originally Posted by amirm View Post
Masking applies to everyone. I thought you were saying that you fail the tests in this thread including your own because you simply can't hear high frequencies above 8 Khz.
I didn't say that I can't hear high frequencies above 8 KHz, and its not true. I can hear pure tones at reasonable levels up to about 12 KHz. But that is at least 7 KHz worse than my hearing was 10 years ago.

General rule of thumb is that you need to be able to hear most if not all of the next critical band in order to detect its absence. The center frequencies of standard critical bands for > 8 KHz do not show up in any of the references I can find, so I'll approximate them as 1/3 octave bands which run like: 8, 10, 13, 16, 20 Khz. That means that the 10 KHz band is the highest one I hear fully, and I should be able to hear an 8 KHz brick wall filter in a DBT, which I have accomplished since my last report.

Quote:
Originally Posted by amirm View Post
If you are now saying that this is due to masking, how come masking didn't stop me and others from hearing the degradations but did that to you???
Your ears are better than mine which means that you can hear more high frequency critical bands, etc. Things scale.
arnyk is offline  
Old 07-20-2014, 10:01 PM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by koturban View Post
I replied to your post within its context. ie, what you posted. It's disappointing, but not the least surprising, that you continue to evade even the most straightforward questions.
Next to disappointing Arny, disappointing you is one of my biggest fears in life.

Quote:
Originally Posted by kot
Who is this "we"?
The three of us in this picture:



And oh, Chu would have also been in that picture but was sick that day.

Quote:
Originally Posted by kot
Anecdote.


Quote:
Originally Posted by kat
That, and a dollar will buy a cup of coffee.
Story of my life....

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-20-2014, 10:06 PM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by arnyk View Post
...I should be able to hear an 8 KHz brick wall filter in a DBT, which I have accomplished since my last report.
Quote:
Originally Posted by arnyk View Post
I now struggle to hear the effects of an 8 KHz brick wall filter at normal listening levels.
You can pass a DBT while struggling to hear what is being tested???

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-20-2014, 10:25 PM
AVS Addicted Member
 
Chu Gai's Avatar
 
Join Date: Sep 2002
Location: NYC area
Posts: 14,968
Mentioned: 1 Post(s)
Tagged: 0 Thread(s)
Quoted: 318 Post(s)
Liked: 596
Quote:
Originally Posted by amirm View Post

What do you think of this as a teaching aid for masking?

I doubt it's better than this.

tubetwister likes this.

"I've found that when you want to know the truth about someone that someone is probably the last person you should ask." - Gregory House
Chu Gai is online now  
Old 07-21-2014, 12:06 AM
AVS Special Member
 
NorthSky's Avatar
 
Join Date: Jan 2013
Location: Canada - West Island: Vancouver, South Direction: Go East
Posts: 4,934
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 2134 Post(s)
Liked: 612
This thread is getting cuter and cuter by the moment.
NorthSky is offline  
Old 07-21-2014, 12:25 AM
Newbie
 
onlyconnect's Avatar
 
Join Date: Nov 2011
Posts: 5
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 7 Post(s)
Liked: 2
Quote:
Originally Posted by arnyk View Post
In addition downsampling should be pre-filtered, then downsampled and post-filtered in order to avoid aliasing. The very minor timing and frequency response variations caused by the pre and post filtering again makes it impossible for the files to null perfectly.
Well, I assume all those downsampling options in Audition are there for a good reason. The question though: is Amir testing the downsampling algorithm, or redbook vs hi-res?


Tim
onlyconnect is offline  
Old 07-21-2014, 12:49 AM
AVS Special Member
 
RayDunzl's Avatar
 
Join Date: Sep 2013
Location: Tampa, Florida
Posts: 1,403
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 480 Post(s)
Liked: 679
It's become a test of whatever the Naysayers say "nobody can hear" and then Amirm tries it and replies with "Oh yes I can..."

I'll be back later...


links::: 1.5RQ > digits > 1177a > OpenDRC-DI > DEQ2496 > DAC2 > KCT > FPB 350mcx > reQuest > Sweetspot
RayDunzl is offline  
Old 07-21-2014, 01:25 AM
Member
 
rufus13's Avatar
 
Join Date: Dec 2012
Location: Soviet of Portlandia
Posts: 20
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 11 Post(s)
Liked: 4
Quote:
Originally Posted by RayDunzl View Post
It's become a test of whatever the Naysayers say "nobody can hear" and then Amirm tries it and replies with "Oh yes I can..."
Then, anyone can hear it on headphones, but not on favorite no-WAF loudspeaker. I still think that this is due to non-optimized rooms in housing (it's housing, not studio from ground-up), as well as consumer loudspeakers tending to make program sound "good" rather than brutally accurate (no one wants "warts and all" sound, even if they say they do, unless being paid hourly for it).

Asserting a negative means that comprehensive and significant evidence must be presented. Then there's the guy who says: "I did/saw/made/photographed/recorded that thing you said was impossible, and here's how"

I propose that if anyone can detect any effect or artifact, by any repeatable means, the thing exists. Even better, tell us what's wrong and how to fix it. Alternate explanation? Swamp gas, mass hallucination, or Aurora hypersonic craft?

Is it significant, when included in actual program, instead of in a test signal? That depends on half a dozen moving parts, not least of being the listener.
rufus13 is offline  
Old 07-21-2014, 01:31 AM
Member
 
rufus13's Avatar
 
Join Date: Dec 2012
Location: Soviet of Portlandia
Posts: 20
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 11 Post(s)
Liked: 4
Quote:
Originally Posted by krabapple View Post
No, not necessarily. Your loudspeakers *might* be fine.



Yes, quite possibly, though 'perfection' is neither necessary or even attainable, except *perhaps* for one very narrow listening position.
A persons' loudspeakers probably are fine. If the music they play back sounds acceptably like what they hear in a concert hall, they are fine.

Most domestic hi-fi situations can be improved a bunch with $1000 in room acoustic treatment/consulting, much more so than by spending a second $10K on system hardware, imho.
rufus13 is offline  
Old 07-21-2014, 02:43 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by onlyconnect View Post
Well, I assume all those downsampling options in Audition are there for a good reason.
Frankly, the dithering and filtering options are there so that they can be used to obtain the cleanest possible results. They have switches on them in case someone wants to see the bad things that happen when they are not engaged, or work in a context where their goals are accomplished by other means. For example dither could be reasonably skipped during the downsampling if it was previously added by other means.

Again I'm being victimized for trying to do the right thing - following the historic strategy of punishing every good deed.

Quote:
Originally Posted by onlyconnect View Post
The question though: is Amir testing the downsampling algorithm, or redbook vs hi-res?
Neither. The obvious strategy is to be as frustrating as possible to deal with. The resulting frustrated attempts at clarification obfuscate the ongoing stream of technical errors like the idea that its reasonable to decrease sample precision without dither.
arnyk is offline  
Old 07-21-2014, 02:52 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by RayDunzl View Post
It's become a test of whatever the Naysayers say "nobody can hear" and then Amirm tries it and replies with "Oh yes I can..."
I guess it should be pointed out that FOOBAR2000 ABX logs are easy enough to fabricate with a text editor. The samples being compared can be processed before the test so that they become far easier to detect. These tricks could be made more difficult if there was a validation serial number that was part of every report, and checksums were taken of the files being compared.

The issue of IM and other flaws in the monitoring chains remains open.

The results of any test are suspect when they can only be obtained by one person and/or using a specific test environment.

Not saying that any of this is actually happening, but also suggesting some caution should be used and results not be taken as being utterly decisive.
koturban and tubetwister like this.

Last edited by arnyk; 07-21-2014 at 03:15 AM.
arnyk is offline  
Old 07-21-2014, 05:09 AM
Senior Member
 
stereoeditor's Avatar
 
Join Date: Feb 2010
Posts: 343
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 140 Post(s)
Liked: 39
Quote:
Originally Posted by arnyk View Post
The context (if memory serves) is that I saw an early draft of JAs article that involved algebraic subtraction of decoded MP3s from the .wav file used to create them, and pointed out the same problems that everybody else who is knowledgeable about perceptual coding would point out.
Your memory serves you ill, Mr. Krueger, as you never saw an "early draft" of an article of mine on lossy compression.

John Atkinson
Editor, Stereophile
stereoeditor is offline  
Old 07-21-2014, 06:25 AM
Member
 
jkeny's Avatar
 
Join Date: Nov 2007
Posts: 16
Mentioned: 0 Post(s)
Tagged: 0 Thread(s)
Quoted: 1 Post(s)
Liked: 11
Quote:
Originally Posted by arnyk View Post
I guess it should be pointed out that FOOBAR2000 ABX logs are easy enough to fabricate with a text editor. The samples being compared can be processed before the test so that they become far easier to detect. These tricks could be made more difficult if there was a validation serial number that was part of every report, and checksums were taken of the files being compared.

The issue of IM and other flaws in the monitoring chains remains open.

The results of any test are suspect when they can only be obtained by one person and/or using a specific test environment.

Not saying that any of this is actually happening, but also suggesting some caution should be used and results not be taken as being utterly decisive.
Arny, I see you have been reading another thread that is discussing these tests or Onlyconnect has suggested these points to you in PM - they are exactly the points raised on the PFM thread.

There is another possibility - the one that Onlyconnect has already mentioned - your downsampling has introduced some audible distortions which is what actually Amir is hearing as the difference between the files. Your use of an "ancient" SRC is also commented on in that thread as one of the possible reasons for these distortions. I've done some null tests on 1644 files I downsampled from your original 24/96 file. I used R8brain & the latest Audition to go to 16/44 & back to 24/96. The using Audio Diffmaker to compare these 2 files to your 16/44 file I get a 13.53usec offset & a null depth of somewhere around 80dB (it varies slightly between the files) so I'm surmising that your downsampled file is fine but has an offset (no audible penalty here, maybe)

Last edited by jkeny; 07-21-2014 at 06:39 AM.
jkeny is offline  
Old 07-21-2014, 07:07 AM
Senior Writer @ AVS
 
imagic's Avatar
 
Join Date: Dec 2005
Location: Philadelphia, PA
Posts: 6,421
Mentioned: 31 Post(s)
Tagged: 0 Thread(s)
Quoted: 1750 Post(s)
Liked: 3077
Quote:
Originally Posted by arnyk View Post
I guess it should be pointed out that FOOBAR2000 ABX logs are easy enough to fabricate with a text editor. The samples being compared can be processed before the test so that they become far easier to detect. These tricks could be made more difficult if there was a validation serial number that was part of every report, and checksums were taken of the files being compared.

The issue of IM and other flaws in the monitoring chains remains open.

The results of any test are suspect when they can only be obtained by one person and/or using a specific test environment.

Not saying that any of this is actually happening, but also suggesting some caution should be used and results not be taken as being utterly decisive.
Arny, I had all those thoughts regarding Amir's results with the key jangling files. Then, I simply figured out how to pass the test. So, based on personal experience, cheating is totally unnecessary. That's because with those files, there is an obvious, detectable difference, even between 44.1 and 96.
amirm likes this.

Find out more about Mark Henninger at www.imagicdigital.com

Last edited by imagic; 07-21-2014 at 07:10 AM.
imagic is online now  
Old 07-21-2014, 07:09 AM - Thread Starter
AVS Addicted Member
 
amirm's Avatar
 
Join Date: Jan 2002
Location: Washington State
Posts: 18,375
Mentioned: 4 Post(s)
Tagged: 0 Thread(s)
Quoted: 981 Post(s)
Liked: 547
Quote:
Originally Posted by RayDunzl View Post
It's become a test of whatever the Naysayers say "nobody can hear" and then Amirm tries it and replies with "Oh yes I can..."
I don't know that I can do this in all cases but hopefully the evidence is clear that listeners are differentiated based on samples presented in this forum. The results indicate that you can't take some graph from a book and say, "look, you are not supposed to hear this." Well, those graphs didn't include me so by definition they are wrong. And could have assumptions that are not shared in your tests. Or, you are not understanding the graph.

Now the tests could be flawed in which case, shame on people who created them seeing how they are supposed to be experts in creating such tests. Present company excluded of course . Ultimately tests were presented such that the experiment creator could not hear the difference but others could. That clearly establishes the differential in hearing abilities. As you rightly say, it almost doesn't matter what the test is. The fact that the expert put it forward as inaudible yet proved otherwise, makes it clear that some people do hear better others. So one should not use their own hearing ability as judgement of what is or is not audible to others. And certainly not when you admit that filtering everything above 8 Khz is a struggle for you to hear. It is like asking a homeless person investing advice.

Amir
Retired Technology Insider
Founder, Madrona Digital
"Insist on Quality Engineering"
amirm is offline  
Old 07-21-2014, 07:15 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by stereoeditor View Post
Your memory serves you ill, Mr. Krueger, as you never saw an "early draft" of an article of mine on lossy compression.
Well, it wasn't the 2008 article, whose use of charts and graphs reminds me of that classic book: "How to Lie With Statistics", and it wasn't the brief 1999 article. Probably from before the 1998 bankruptcy of SP's parent...
arnyk is offline  
Old 07-21-2014, 07:18 AM
AVS Addicted Member
 
arnyk's Avatar
 
Join Date: Oct 2002
Location: Grosse Pointe Woods, MI
Posts: 14,387
Mentioned: 2 Post(s)
Tagged: 0 Thread(s)
Quoted: 763 Post(s)
Liked: 1178
Quote:
Originally Posted by imagic View Post
Arny, I had all those thoughts regarding Amir's results with the key jangling files. Then, I simply figured out how to pass the test. So, based on personal experience, cheating is totally unnecessary. That's because with those files, there is an obvious, detectable difference, even between 44.1 and 96.
I figured out how to "Pass the test" too, even with my ratty up-to-12 KHz hearing. Merely turn the volume on my laptop up just enough to get barely audible distortion due to the extra content > 20 KHz. ;-)

BTW, who has confirmed your results with completely independent tests?

How many different combinations of digital players and transducers have you duplicated your results with? if its *that* easy, you should be able to rack up an impressive number of retests on different players and transducers in a heartbeat or two...

Last edited by arnyk; 07-21-2014 at 07:22 AM.
arnyk is offline  
 
Thread Tools


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off