Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: 'Normalization' of PCM audio - subjectively benign? (Read 140759 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

'Normalization' of PCM audio - subjectively benign?

Reply #150
I'll take on the dynamic range article if somebody else comments on the sampling article.

EDIT: I replied to that thread, comments on my analysis welcome.



Thanks Axon, perhaps Clint will consider replacing her article with yours?  I suspec though that he'd want some actual data from using your protocol.

 

'Normalization' of PCM audio - subjectively benign?

Reply #151
Yes, all majors in EE requires a knowledge of Fourier transforms.

However I finally recall that Discrete Signal Processing (in my university) is mandatory only for the Telecommunications major. It is optional for all the other majors... and due to its complexity, naturally no one wants to take that satanic course


Where I did my computer engineering course (University of Sherbrooke in Quebec, Canada), all Computer Engineering & Electrical Engineering students had to take a "Communications" course where we would be exposed to Fourrier and more generally to Laplace transforms.  We would study Shanon and Nyquist.  The final project was the simulate, using Matlab, the quantization & sampling process occuring with music @ 16/44.1.  We had to reconstruct the signal and show the difference with the original signal.  How cool is that?  It was very instructive.

'Normalization' of PCM audio - subjectively benign?

Reply #152
I posted a complaint thread about Tham's stuff on Audioholics.  Any takers here for the reply from the editor?

http://forums.audioholics.com/forums/showt...6095#post206095


Quote
Perhaps you can direct the more knowledgable people on those forums to construct a new test sequence that is more real world? We're open to additional input.
__________________
Clint DeBoer
Editor in Chief
Audioholics




Something (a glitch?) made that thread appear invalid for awhile.  But it's still there.

http://forums.audioholics.com/forums/showthread.php?p=206095

'Normalization' of PCM audio - subjectively benign?

Reply #153
The link doesn't work for me.

'Normalization' of PCM audio - subjectively benign?

Reply #154



I can't see the article either.

EDIT: I will save a "J'ACCUSE!" moment for a short while longer. This could get sensitive.


'Normalization' of PCM audio - subjectively benign?

Reply #156
Yeah... it's gone "sensitive".


'Normalization' of PCM audio - subjectively benign?

Reply #158
 It is obvious you know that, unless you normalize at the pick amplitude, the normalization changes the sound. Another question is how audible it is.
If normalization decreases the sample values it introduces spectrum distortions also.
In any case, I think, it does not influence the quantization noise as seems to me the discussion is about.

For the simple case when the quantization noise is considered “white”, and this model is realistic, as it is said in “Discrete Time Signal Processing” by Alan V. Oppenheim and Ronald W. Shafer when “…the signal is sufficiently complex and quantization steps are sufficiently small so that the amplitude of the signal is likely to traverse many steps from sample to sample”,
I think: The quantization noise depends only on the quantizer used and does not depend on the signal.

So, to me, the normalization in the case of increasing the signal does not distort it as long as it stays within the acceptable number range.

'Normalization' of PCM audio - subjectively benign?

Reply #159
Once more into the breach....

Over on AVSForum, someone named 'DulcetTones" cited the 'Mother of Tone' page as an authoritative reference against the idea that we get 'perfect sine waves' from digital audio,

http://www.avsforum.com/avs-vb/showpost.ph...mp;postcount=74


Quote
I might as well point to these again as some seem intent on saying the usual vague facts.
http://www.mother-of-tone.com/cd.htm
Article is correct, no-one showed anything wrong last time. Apart from some vague comments as usual.
In other words those who say you end up with a perfect sine wave are wrong.
It seems for some reason there are a few who decide to talk about CD and say its a sinewave and yet do not understand the significance of the rectangular sinusoidal component.
This is shown in the above link, you only end up with the sine wave after the reconstruction filters work their logic on the rectangular sinusoidal components shown in the URL.
Also the URL is good as it shows some other challenges with sine waves and PCM.


I pointed out that that article had been, um, *critiqued* on HA.org, (pointing to this thread) as being rather skewed.
(I also agreed that we don't get 'perfect' reconstruction of the input signal from any audio repro medium, but digital properly done is the closest by far)

Much hooing and hahing ensues, during which DT claims HA folk don't know what they're talking about. Finally this:

http://www.avsforum.com/avs-vb/showpost.ph...p;postcount=120

Quote
Krab,
Ok to prove which of the two sources we are quoting actually know what they are talking about.
The person you hold in high regard who is basically stating Mother of Tone do not know what they are talking about commented in the forum you hold in high regard that Mother got the Nyquist and Shannon formulas wrong way round.
In fact your guy was quite explicit in saying this was a fundamental mistake. 

Well hate to burst your bubble bud:
1st Part of the equation discussion on Mother.
Quote

Well, what did Nyquist say ?

maximum data rate in a noiseless channel = C = 2*W log base2( L ) bits/sec

* where 2W is 2 times the highest frequency contained in the noiseless channel, and

* where L = number of discrete levels (e.g., binary = two levels, 0 and 1)

As Nyquist seems to have been more interested in data transmission than in high-fidelity, we should not wonder, that his statement just defines a maxiumum data-rate of a communications channel.


This is EXACTLY CORRECT and pertains to the original Nyquist paper I suggested you read.
http://www.nctt.org/pages/resources/simulations/nyquist.php

Nice simple comparison, exact formula.

Then the next part in Mother expanding on the original works of Shannon:

Quote
Later, Claude Shannon said:

If a function s(x) has a Fourier transform F[s(x)] = S(f) = 0 for |f| > W, then it is completely determined by giving the value of the function at a series of points spaced 1/(2W) apart. The values sn = s(n/(2W)) are called the samples of s(x).

This goes much further than Nyquist's words, in that it states, that a signal which consists of sine waves with a maximum frequency of W is completely described by recording its values twice as fast as W.

The real cool thing is that Shannon also gave a interpolation formula to get back to the original signal:


Unfortunately this formula includes an infinite sum... What does that mean ?


Again this can be seen from a link that just focuses on the data from Nyquist and Shannon.
http://www.fact-index.com/n/ny/nyquist_sha...ng_theorem.html

Save the effort of reading it all here it is:

Quote
If a function s(x) has a Fourier transform F[s(x)] = S(f) = 0 for |f| > W, then it is completely determined by giving the value of the function at a series of points spaced 1/(2W) apart. The values sn = s(n/(2W)) are called the samples of s(x).
Also the Mother of Tone went on to show the Interpolation formula (that does not copy into here) and this matches again EXACTLY the paper by Shannon (can find it in Section II The Sampling Theorem).


So the guy your quoting actually made what I would call FUBAR statement in his argument against Mother of Tone.

The person you quote argues just as persistently as you do, makes mistakes such as stating the "infinite filter" are already here due to filters having 10000s of taps, this is rather obtuse and why I asked you earlier do you know which products are using 256 or 1024 tap lengths.

Notice 1024 is substantially smaller than the 10000s your colleague states and the 1024 type are only found in the significantly more expensive DAC-CD players.

Now why the heck would I want to wade into the bull on another forum when it is giving me a headache arguing with someone even on this forum who does not even take the time to read and follow up on both :
Nyquist Certain Topics in Telegraph Transmission Theory
Shannon Communication in the Presence of Noise



(NB: DT appears to have conflated what MikeG wrote about 'Mother of Tone' getting Shannon and Nyquist crossed, with what 2bdecided said about 'several thousands' (not '10000s') of taps....conflating at least two people, or possibly three if SebastianG's objections are included, into one)

Anyway, I suggested that rather than me arguing by proxy for those on HA that he considers incompetent and wrong about the 'Mother of Tone' article, that he address you directly here, but as you see he demurs.  So I'm bringin' it here, for said parties to respond ...or not.

Btw, I of course have no issue with either Shannon or Nyquist, and my issue on that AVSF thread was with people pointing to the dreaded 'stairsteps' views as an indication of how supposedly 'problematic' Redbook is at various frequencies, and also with the lack of audibility test data from people who expect DACs that measure differently, to routinely sound different.

EDIT: fixed broken URLs in the quote

'Normalization' of PCM audio - subjectively benign?

Reply #160
As I understand it DT acknowledges the sampling theorem and goes on to say that ideal reconstruction is impossible because the ideal reconstruction filter's impulse response (normalized sinc curve) extends infinitly to both sides. So far, so good. But in practice this is not an issue. You can always design a practical reconstruction filter that is good enough. Take CDDA for example: it's not a problem to design a practical filter that faithfully reconstructs the audible band (0-20 kHz) and rejects image frequencies (those above 22.05 kHz) by 100 dB or more. The impulse response of a filter like this can be as short as 2 milliseconds:
[a href="http://img158.imageshack.us/my.php?image=reconstruct44uq7.png" target="_blank"]

Cheers,
SG

'Normalization' of PCM audio - subjectively benign?

Reply #161
A heads up:

uart's 'scope screen cap on this thread. has been referenced (negatively) in a reply by Dr. Milind Kunchur to critiques (from Woodinville on Stereophile's forum, and, apparently, from other critics on other forums)  of his published claims about the insufficient 'time resolution' of Redbook, briefly summarized here

http://www.physics.sc.edu/kunchur/Acoustics-papers.htm

Woodinville's (and Axon's)  initial and new critiques are here

http://forum.stereophile.com/forum/showfla...age=0#Post69464

Dr. Kunchur's reply is here:

http://www.physics.sc.edu/kunchur/papers/FAQs.pdf

'Normalization' of PCM audio - subjectively benign?

Reply #162
A heads up:

uart's 'scope screen cap on this thread. has been referenced (negatively) in a reply by Dr. Milind Kunchur to critiques (from Woodinville on Stereophile's forum, and, apparently, from other critics on other forums)  of his published claims about the insufficient 'time resolution' of Redbook, briefly summarized here

http://www.physics.sc.edu/kunchur/Acoustics-papers.htm

Woodinville's (and Axon's)  initial and new critiques are here

http://forum.stereophile.com/forum/showfla...age=0#Post69464

Dr. Kunchur's reply is here:

http://www.physics.sc.edu/kunchur/papers/FAQs.pdf


The equivocation and attempt to deflect is really pretty astonishing to me.

The unwillingness to actually address the facts of the frequency response difference implied by a two-tap FIR filter with the two taps 5 microseconds apart is, for me, the final straw. Referring to a mechanism that creates an audible difference within the 20kHz band as mathematical nitpicking is simply not a reasonable response.
-----
J. D. (jj) Johnston

'Normalization' of PCM audio - subjectively benign?

Reply #163
A heads up:

uart's 'scope screen cap on this thread. has been referenced (negatively) in a reply by Dr. Milind Kunchur to critiques (from Woodinville on Stereophile's forum, and, apparently, from other critics on other forums)  of his published claims about the insufficient 'time resolution' of Redbook, briefly summarized here

http://www.physics.sc.edu/kunchur/Acoustics-papers.htm

Woodinville's (and Axon's)  initial and new critiques are here

http://forum.stereophile.com/forum/showfla...age=0#Post69464

Dr. Kunchur's reply is here:

http://www.physics.sc.edu/kunchur/papers/FAQs.pdf


That's interesting, I hadn't seen that pdf-file reply before. Just one quick question, in the reply he repeatedly makes reference to the "5 microsecond or better temporal resolution of the human hearing". Does anyone know where he got that figure from or how that was measured? It seems a bit extreme to me.

'Normalization' of PCM audio - subjectively benign?

Reply #164
He measured it himself, in some (peer-reviewed!) papers, and actually, it's a figure that nobody involved (jj or krab or I) really disagree with. What this whole spat is about is the meaning of such a result, especially in the context of the fidelity of redbook, and in a larger context, the meaning of the term "resolution".

There's a really long backstory/flamewar behind all of this that you have not been introduced to (and I won't even try to do so). krab only posted because I used your scope shots as a trivial demonstration of the ability of mainstream sound cards to reproduce 14khz sine waves without distortion. Kunchur's response to your plot is a little bizarre to me, but I'm still crunching on everything before I will comment any further.

'Normalization' of PCM audio - subjectively benign?

Reply #165
He measured it himself, in some (peer-reviewed!) papers, and actually, it's a figure that nobody involved (jj or krab or I) really disagree with. What this whole spat is about is the meaning of such a result, especially in the context of the fidelity of redbook, and in a larger context, the meaning of the term "resolution".

There's a really long backstory/flamewar behind all of this that you have not been introduced to (and I won't even try to do so). krab only posted because I used your scope shots as a trivial demonstration of the ability of mainstream sound cards to reproduce 14khz sine waves without distortion. Kunchur's response to your plot is a little bizarre to me, but I'm still crunching on everything before I will comment any further.


Specifically, he plays one pulse, and then two half-amplitude pulses separated by 5 microseconds, and listeners with unimpaired hearing in good settings can hear the difference.

Given that this is a comb filter with its first zero at 100kHz, being able to hear the difference at 20kHz isn't too surprising, now.

So, this shows a mechanism for which purely in-band (i.e. dc - 20kHz) signal can allow the distinction.

Let's see....

Down about .45dB is just at the DL.  In really good conditions, for direct, immediate comparison, with no level roving, etc.

So, what did the doctor actually discover?

Nothing that I can tell.

This, of course, after misdirection upon misdirection from the people 'defending' his work, starting with quotes that made things look like it was interaural resolution (which is also audible, just barely, in the best conditions, at that level), defending his claim that you couldn't get that kind of time resolution out of a PCM signal, rejecting plots showing otherwise, changing the claims and the goalposts repeatedly and at the same time accusing the people trying to have a dialog of "shifting their excuses", and repeated, exhaustive references to Dr. K's PhD and how publishing this in a conference record was the full endorsement of the organization, etc.

The people who are saying this do, I think, know better, too.
-----
J. D. (jj) Johnston

'Normalization' of PCM audio - subjectively benign?

Reply #166
Oh, and the best joke of all this was that the intentionally ignorant over there think I was being rude.

Oh, baby, they have no idea what RUDE is. Apparently Gauss, Fourier, Fletcher and Zwicker are also rude, in any case.

It's just like creationists, they only accept confirmatory data.
-----
J. D. (jj) Johnston

'Normalization' of PCM audio - subjectively benign?

Reply #167
It would be fun to repeat the experiment, but hook a CD recorder into the loop, testing if (proving that!) the "effect" was still audible after passing through 44.1kHz / 16-bits.

Hopefully that would be simple enough to show how silly this is.

Cheers,
David.

'Normalization' of PCM audio - subjectively benign?

Reply #168
It would be fun to repeat the experiment, but hook a CD recorder into the loop, testing if (proving that!) the "effect" was still audible after passing through 44.1kHz / 16-bits.

Not only fun, but essential to prove his claim that CDDA is not good enough, and furthermore a test that his recommended 192kHz is the way to go. Without that, where is the relation between his artificial laboratory tests (5 years long and so very PhD-like theoretical), and reality of music reproduction?

Prof. Kunchur talks about the big picture which his results have to be viewed within. He himself writes in his FAQ about errors that are acceptable in consumer audio, but not in his experiment. So even if he found an error in CDDA, is it reason enough to replace it? He says he is an audiophile, and that's the reason his answer is yes. I come to this conclusion because he has "experienced hearing loss ... from riding sports motorcycles", listens to cassettes and LPs, but on the other hand was not happy with the sound of his CD transport/DAC combination, so he needed a "jitter buster" appliance. So the big picture for me is biased research which is funded by those having an interest in a certain result.

'Normalization' of PCM audio - subjectively benign?

Reply #169
He says he is an audiophile, and that's the reason his answer is yes. I come to this conclusion because he has "experienced hearing loss ... from riding sports motorcycles", listens to cassettes and LPs, but on the other hand was not happy with the sound of his CD transport/DAC combination, so he needed a "jitter buster" appliance. So the big picture for me is biased research which is funded by those having an interest in a certain result.


Wait, Dr. Kunchur funded his own research??  Nice for those who can!

Actually the audio-related research reported in his three journal publications appears to have been funded (well, "partially') intramurally  -- i.e., by his own university.  I don't see any indication of industry bias.  This is in contrast to the vast evil CONSPIRACY involving the AES, the low-end audio industry, Hydrogenaudio, the Freemasons, and our secret alien lizard overlords, as uncovered by certain brave Stereophile forum posters!     


'Normalization' of PCM audio - subjectively benign?

Reply #170
Specifically, he plays one pulse, and then two half-amplitude pulses separated by 5 microseconds, and listeners with unimpaired hearing in good settings can hear the difference. Given that this is a comb filter with its first zero at 100kHz, being able to hear the difference at 20kHz isn't too surprising, now. So, this shows a mechanism for which purely in-band (i.e. dc - 20kHz) signal can allow the distinction. Let's see.... Down about .45dB is just at the DL.  In really good conditions, for direct, immediate comparison, with no level roving, etc. So, what did the doctor actually discover? Nothing that I can tell.
Um, no?

His experiments only involve pulses by implication - ie, that playing two time-misaligned sine waves is equivalent to one sine wave plus two rectangular pulses, or that a first-order-filtered square wave can be roughly treated as a square wave plus two exponentially decaying pulses. Neither of which actually involve pulses spaced 5us apart. Granted, everybody involved (including Dr. Kunchur) seem to be all too happy to reduce this down to the spaced pulses idea, but let's keep it real on what is actually in the papers.

Combing does occur with his speaker-alignment experiment but he goes to lengths to measure the headphone output and assert that the DL is not reached at any frequency, and his method appears more or less satisfactory to me (at least as far as this particular part of the procedure is concerned).

Quote
This, of course, after misdirection upon misdirection from the people 'defending' his work
Yeah, there's no excuse for that, but seriously, what did you expect out of a gaggle of audiophiles like that? It's not all that productive to drag sasaudio etal into a fight that really should just be between you and Kunchur.

Oh, and the best joke of all this was that the intentionally ignorant over there think I was being rude. Oh, baby, they have no idea what RUDE is. Apparently Gauss, Fourier, Fletcher and Zwicker are also rude, in any case. It's just like creationists, they only accept confirmatory data.
The impression I got was that Fourier (at the least) was a gasbag of a reasonably high order. I would not call him a role model.

It would be fun to repeat the experiment, but hook a CD recorder into the loop, testing if (proving that!) the "effect" was still audible after passing through 44.1kHz / 16-bits. Hopefully that would be simple enough to show how silly this is.
Curiously enough Kunchur never does run this control (ie, run the tests with 7khz sine instead of 7khz square, and/or 7+21khz sines). I'm unsure how important this is.

He says he is an audiophile, and that's the reason his answer is yes. I come to this conclusion because he has "experienced hearing loss ... from riding sports motorcycles", listens to cassettes and LPs, but on the other hand was not happy with the sound of his CD transport/DAC combination, so he needed a "jitter buster" appliance. So the big picture for me is biased research which is funded by those having an interest in a certain result.


Wait, Dr. Kunchur funded his own research??  Nice for those who can!

Actually the audio-related research reported in his three journal publications appears to have been funded (well, "partially') intramurally  -- i.e., by his own university.  I don't see any indication of industry bias.  This is in contrast to the vast evil CONSPIRACY involving the AES, the low-end audio industry, Hydrogenaudio, the Freemasons, and our secret alien lizard overlords, as uncovered by certain brave Stereophile forum posters!     
Also note that Hi-Fi Critic did a feature on Kunchur (as an example of one of the "good guys", meaning somebody who justifies/apologizes for high end audio).

There's an extremely critical point to be made related to all of this, but I'm saving it to address to the good doctor himself.

'Normalization' of PCM audio - subjectively benign?

Reply #171
Prof. Kunchur ... listens to cassettes and LPs, but on the other hand was not happy with the sound of his CD transport/DAC combination, so he needed a "jitter buster" appliance.

That says an awful lot IMO.

I wonder if he has any bass traps and other room treatment.

--Ethan
I believe in Truth, Justice, and the Scientific Method