Conclusive "Proof" that higher resolution audio sounds different

[jkeny]

Well I got positive results too using Arny's files. But it looks like a poor resampler was used. Of course the files are quite old and software resampling has improved. When I resampled the full bandwidth file with Audacity I could no longer pick a difference in Foobar abx testing. You might try that yourself, and see if it is still audible to you. Not that such is a debunking. But the quality of resampling is what was audible to me at least. When done better it was no longer audible. You might still pick out a difference however.

Have you any evidence that the quality of the original resampling was bad? Some null testing or identification of audible spectrum or timing differences between old & new resampling that you can show?

 

[jkeny]

Totally agree of course but what would Data say after he's been succumbed by the "Borg"?

He would say whatever the collective mind wants him to say

 

[still-one]

Totally agree of course but what would Data say after he's been succumbed by the "Borg"?

I have no idea what you are referring to.

 

[Phelonious Ponk]

I can't understand the logic of someone who says "hearing is not to be trusted" because the brain is easily influenced & yet says that JUST taking away knowledge of what's being listened to is revealing the truth about what we hear? It must surely be realised that the brain is still active (& influenced) when listening blind. It must surely be realised that there are many factors still affecting the listening perception. It must surely be realised that sticking electrodes on the auditory nerve coming from the ear will not tell us much about what we perceive - doing the same on the optic nerve will tell us little about our perception of vision. It's the processing of the information that is critical.

As I said already, fully controlled blind listening tests are an attempt to grapple with the known factors that influence our hearing perception. Ignoring these factors or doing a half-arsed job & claiming some higher level of truth is just a pretense & one that hampers discovery.

My sig says this better than I "The greatest obstacle to discovery is not ignorance – it is the illusion of knowledge." - Daniel J. Boorstin.

Good line. I hope you keep it handy.

Tim

 

[microstrip]

Good line. I hope you keep it handy.

Tim

Those wanting to know the exact quote and how internet supports "The illusion of knowledge" can take a look in this page http://backreaction.blogspot.pt/2008/05/illusion-of-knowledge.html

 

[esldude]

Have you any evidence that the quality of the original resampling was bad? Some null testing or identification of audible spectrum or timing differences between old & new resampling that you can show?

Yes. The first pair of images are from Arny's files and the second pair from my resampling Arny's full bandwith 24/96 file using shaped dither.

Along with the FFT of the entire file you also will see the waveform view of the first 5 seconds. The FFT is the nulled out difference. The waveform view is the residual file from nulling with a steep 20 khz filter applied to remove the ultrasonic info which is still in the residual. The residual from my resampling procedure produces a straight line (yes I know if looked at in detail it isn't a full zero waveform). The other shows residual signal left after nulling Arny's files. If you listen to this whether before or after applying the 20 khz filter you hear faintly jangling keys. In the file I resampled you hear nothing. If I amplify greatly the residual of my resampling after applying the 20khz filter, you eventually hear hiss. Mostly an even whitish sounding noise. It is slightly modulated at times though never anything resembling keys or anything identifiable.

I would have thought the residual in Arny's files would have been inaudible, but that proved not to be the case as I could ABX it successfully. The files I resampled in Audacity I could not identify.

 

[sasully]

So has anyone ever successfully differentiated two (or more) DACs in a blind test before?

Sure.

http://home.provide.net/~djcarlst/abx_cd.htm

 

[sasully]

In this forum, we discuss what is best. Not what is good enough for "most listeners."

A stance which could quickly lead to nonsense , depending on how 'best' is defined. E.g., if 'best' means the format that most accurately captures the highest frequencies at the deepest bitdepth, the 'best' , with current tech, would be 192kHz/24bit. Now go peddle that claim to, say, Dan Lavry or Bruno Putzys and see what they say.


It is an ideal way to define 'best' for those peddling 'audiophile' hardware, though. Whatever measures best is best! (Except when it's vinyl, of course. Vinyl is always best. )

 

[Ronm1]

I have no idea what you are referring to.

Aaah! But a Trekkie would know

 

[sasully]

So we see what aggregation of general public does in these tests and why industry recommendation is to use only trained listeners. Inclusion of large number of testers without any prequalifications pushes the results to 50-50. Imagine 4 out of his 140 being like me. Their results would have been erased by throwing them into the larger pool. If our goal to say what that larger group can do, then this kind of averaging of the results is fine. But if we want to make a "scientific" statement about audibility of 16 vs 24, is wrong and we run foul of simpson's paradox. See: http://en.wikipedia.org/wiki/Simpson's_paradox.

As always, I want to caution people that my testing is all about finding a difference and not stating what is better. And yet again, I do not know which file was which as I did my testing. I simply characterised the difference between A and B clips and then went to town.

Of note, I don't know how he had determined that the original files did indeed have better dynamic range than 16 bits. Maybe someone less lazy than me can find it and tell me .

Here is the original thread where the files were posted: http://archimago.blogspot.ca/2014/06/24-bit-vs-16-bit-audio-test-part-i.html

And results: http://archimago.blogspot.com/2014/06/24-bit-vs-16-bit-audio-test-part-ii.html

Brilliant. Now all you need to do is provide evidence that the shock troops of the high rez campaign - from elite high end reviewers at Stereophile or TAS or computeraudiophile, the high-end boutique gear sellers, through the middlebrows like Sound & Vision reviewers through the online swooners on forums like this, SA-CD.net,etc - are *trained listeners* and *really are* hearing the je ne sais quoi that hi rez imparts, rather than artifacts...or no difference at all...when they listen to hi rez music.


Btw, are you sure that Meyer and Moran did not do any per-subject analysis of their data to find the golden ears? Because the paper, which I'm sure you have read, seems to suggest otherwise:

As the tests progressed, we repeatedly sorted the data
for correlations with age, sex, upper frequency hearing
limit, or experience. No such correlations have emerged.
Specifically, on music at normal levels as defined here,
audiophiles and/or working recording-studio engineers got
246 correct answers in 467 trials, for 52.7% correct. Females
got 18 in 48, for 37.5% correct. Those subjects able
to hear tones above 15 kHz got 116 in 256 trials, for 45.3%
correct; listeners aged 14–25 years old (who were, as it
turned out, the same group), also got 116 correct in 256
trials, 45.3%. The “best” listener score, achieved one
single time, was 8 for 10, still short of the desired 95%
confidence level. There were two 7/10 results. All other
trial totals were worse than 70% correct.

Their hundreds of subjects, as you also surely know, since you've read the paper, ranged from students to self professed audiophiles to audio professionals. Alas, apparently not a 'trained listener' amongst them. Though they did of course use music in these trials, not Arny's jangly keys file.

 

[Whatmore]

I am not a "trained listener", nor do I have anything close to good ears. In fact my wife constantly complains that I'm going deaf and on those "test how high you can hear" tests my kids are complaining all the way up to 20khz whereas I'm lost after about 16.

As much as I'd love to have golden ears, it ain't so. That's why I suspect there's a fault in the files.
That and the fact that I got 100% on my laptop speakers

 

[maxflinn]

As I said already, fully controlled blind listening tests are an attempt to grapple with the known factors that influence our hearing perception. Ignoring these factors or doing a half-arsed job & claiming some higher level of truth is just a pretense & one that hampers discovery.

John, something you've also said already - 'long-term sighted comparisons can reveal audible differences not revealed by controlled ABX testing'.

(I've taken the quote from the Pink Fish forum)

Is this a faith-based belief? If not, how do you explain it given your insistence that any test that doesn't account for every conceivable variable is worthless?

I think there's a clear contradiction here that needs clearing up.

 

[jkeny]

John, something you've also said already - 'long-term sighted comparisons can reveal audible differences not revealed by controlled ABX testing'.

(I've taken the quote from the Pink Fish forum)

Is this a faith-based belief? If not, how do you explain it given your insistence that any test that doesn't account for every conceivable variable is worthless?

I think there's a clear contradiction here that needs clearing up.

Max, I think I've explained it already - no, I don't have evidence that long term listening reveals more differences than improperly constructed blind testing - to gather such evidence would require longitudinal studies which I doubt have been done.

I've said that improperly controlled blind tests are no better & probably worse than sighted tests - doesn't make them worthless but does put them in a different light than the light they are usually presented in.

 

[jkeny]

View attachment 16757View attachment 16758View attachment 16759View attachment 16760

Yes. The first pair of images are from Arny's files and the second pair from my resampling Arny's full bandwith 24/96 file using shaped dither.

Along with the FFT of the entire file you also will see the waveform view of the first 5 seconds. The FFT is the nulled out difference. The waveform view is the residual file from nulling with a steep 20 khz filter applied to remove the ultrasonic info which is still in the residual. The residual from my resampling procedure produces a straight line (yes I know if looked at in detail it isn't a full zero waveform). The other shows residual signal left after nulling Arny's files. If you listen to this whether before or after applying the 20 khz filter you hear faintly jangling keys. In the file I resampled you hear nothing. If I amplify greatly the residual of my resampling after applying the 20khz filter, you eventually hear hiss. Mostly an even whitish sounding noise. It is slightly modulated at times though never anything resembling keys or anything identifiable.

I would have thought the residual in Arny's files would have been inaudible, but that proved not to be the case as I could ABX it successfully. The files I resampled in Audacity I could not identify.

Thanks esldude
Is this difference between the two ArnyK files possibly due to clock drift?
No clock drift in your resampled files because you are doing the analysis on the same computer as you did the downsampling.

 

[esldude]

Thanks esldude
Is this difference between the two ArnyK files possibly due to clock drift?
No clock drift in your resampled files because you are doing the analysis on the same computer as you did the downsampling.

There doesn't appear to be a drift. There may be a tiny sub-sample time shift though I think not. Maybe some resampler software also has that side effect. Sox doesn't shift timing or level.

A drift would not be due to different computers unless you sent the resampled file through an AD/DA stage. If the AD and DA were using separate clocks you would get a timing drift between them. If the AD and DA used the same clock or were clock locked you would get a simple timing shift from source file to copy. If you used a one meter cable for instance the shift would be roughly in the range of a nanosecond which would show up in the residuals quite clearly. Clock drift in the items I have checked it on usually is in the low parts per million range.

It was my understanding Arny did his resampling digitally so changing computers should have no effect. Changing software however would.

 

[Orb]

All good points esldude.
I may be wrong but if I remember correctly he did run his own rsampling at one stage and redid the ABX (still passed).
But his context (appreciate this is definitely not your point, just bring it back around to the thread) is not whether A sounds better than B, but rather it makes sense to have the source that is not "messed" around with due to potential transparency challenges that is being heard; and considering most files are recorded these days in at least 24bit and 96khz then makes sense to have these.
Of course my POV beyond this is that hirez (with higher sampling) also means greater flexibility in using slow rolloff/minimal phase/filter design combining aspects of minimum-linear.
It would need investigating but I find it interesting reviewers listening to MBL's C31 CD-DAC sounds more "warm-rich" when compared to the Noble DAC they also reviewed, and the Noble DAC (not talking about the 2014 model) when measured at the output has more of a rolloff and subtly more distortion.
And I agree, the MBL C31 does seem unusual (not a bad thing btw) in terms of its sound compared to its traditional measurements; from what I understand this is using some bespoke combination of minimum-linear phase design.
Apologies for digressing with the anecdotal just putting forward why I feel hirez can be important vs "CD" due to filter designs and possibly their perceived effect.

Back on topic, however I am not sure which files Amir did his own resampling on.
Thanks
Orb

 

[Ronm1]

Changing software however would.

That's it, its a S/W problem not a H/W one. For awhile I thought it might have been my hearing reading this thread.

 

[jkeny]

There doesn't appear to be a drift. There may be a tiny sub-sample time shift though I think not. Maybe some resampler software also has that side effect. Sox doesn't shift timing or level.

A drift would not be due to different computers unless you sent the resampled file through an AD/DA stage. If the AD and DA were using separate clocks you would get a timing drift between them. If the AD and DA used the same clock or were clock locked you would get a simple timing shift from source file to copy. If you used a one meter cable for instance the shift would be roughly in the range of a nanosecond which would show up in the residuals quite clearly. Clock drift in the items I have checked it on usually is in the low parts per million range.

It was my understanding Arny did his resampling digitally so changing computers should have no effect. Changing software however would.

I posted this in reply to you earlier in the thread
"Yes, running through Audio Diffmaker shows an offset of 13.53usec between Arny's 24/96 & the downsampled tracks. This is the equivalent of about half a sample - I doubt that it's audible although I do know that the just noticeable Interaural time difference is around 4usecs. But what we're talking about here is a shift of both stereo channels by 13.53usecs, not shifting one of the stereo channels. "

"- one other thing is that there seems to be timing drift between the files which if not turned on in Diffmaker gives a much worse null of -30dB or so"

Do both of these factors explain the difference in the graphs that you have shown?
Would these differences would be audible?
- a half sample offset of the resampled file
- a timing drift between the two files

Anyone can prove this for themselves:
- download Dfiffmaker
- run a difference between ArnyK's 24/96 file & his downsampled 16/44 file
- do it with different settings in Diffmaker "enable time alignment" turned on & again with it turned off
- compare the nulling values between these two runs

For further checking
- There are intermediate files created by Diffmaker which are the aligned versions of the 16/44 file
- bring that into Adobe Audition & null it against the original 24/96 file
- you will see that if you skip the first 0.187 seconds of the file (some Diffmaker glitch in the aligned file), the nulling is around -100dB for all frequencies <20KHz
- this might be about 10dB worse than the nulling you get when you run resampling using a modern resampler
- Did ArnyK's report that there is a 0.2dB difference between the original file & downsampled files & would this explain the 10dB null diff?

 

[Orb]

That's it, its a S/W problem not a H/W one. For awhile I thought it might have been my hearing reading this thread.

But it can also be H/W one when considering the complete chain at both studio and also at home; both can "mess" around with source file.
One aspect still not ruled out is the dither that a few of us went into a bit detail earlier on in this thread (including aspects of TPDF and other types, when used-implications,etc).
Esldude out of curiosity which dither did you use when downsampling-decimation (sampling rate and bit depth)?

Thanks
Orb

 

[maxflinn]

Max, I don't have evidence that long term listening reveals more differences than improperly constructed blind testing

John, you said 'long-term sighted comparisons can reveal audible differences not revealed by controlled ABX testing'

Not - long term listening reveals more differences than improperly constructed blind testing

I've said that improperly controlled blind tests are no better & probably worse than sighted tests - doesn't make them worthless but does put them in a different light than the light they are usually presented in.

But clearly the more variables you remove, the more valid the test. Why then would sighted tests be better than any form of blind tests in terms of identifying differences?