more detail on how Kunchur "sorted his ABX"?
as I recall he kept saying naïve things about digital audio's ability to place "audio events" relative to each other in time - its nanoseconds at 16/44 and >80 dB dynamic range
then he created a mechanical setup that would be expected to give substantial audible errors due to radiation directivity interaction from the transducer being moved back and forth
as I recall he kept saying naïve things about digital audio's ability to place "audio events" relative to each other in time - its nanoseconds at 16/44 and >80 dB dynamic range
then he created a mechanical setup that would be expected to give substantial audible errors due to radiation directivity interaction from the transducer being moved back and forth
Last edited:
You were talking about "differences" and I asked you what kind differences you are referring to that's highlighted in red below.
I must be missing something here because Dave has explained what he's talking about about as clearly as it's possible to do so!
Best Regards,
TerryO
What you don't seem to know that Clark's tests are for the most part my tests. We were partners "in crime". ;-) Read his AES paper and you will find my name.
Been there, done that. Kunchar's false claims about digital timing are widely debunked. JCX already outlined their fatal flaw a few posts back.
I may have been the first to point out this flaw on a public forum - it was the Stereophile forum. Interesting papers. | Stereophile.com
His test was no better than his false theories about digital. What he proved is hard to describe without the description being open to misinterpretation as an insult.
Making things more complex than they need to be is a well-known tactic that is part of deflection. After their logical arguments are crushed it is not uncommon for people to add unecessary complexity or deflections to conceal their loss.
There is a nearly unending list of tricks that one can use to appear to be irrefutable. One really classic one is over a century old:
http://http://www.mnei.nl/schopenhauer/38-stratagems.htm
Actually JCX is right on.
Doesn't help - the gross errors in the paper can't be fixed by simply rereading it. A rewrite would be in order but that would be Kunchur admitting he was wrong...
May i suggest rereading of his papers and his additional PDF but maybe some good texts on digital audio theory before that?
Maybe, with adequate preparation, you'll see the gross errors this time.
Let´s see:
(Source: Milind Kunchur, FAQs.PDF, Seite 1; http://boson.physics.sc.edu/~kunchur/papers/FAQs.pdf )
Last sentence emphasized by me.
(Source: Milind Kunchur, FAQs.PDF, Seite 1; http://boson.physics.sc.edu/~kunchur/papers/FAQs.pdf )
Last sentence emphasized by me.
Oh my Lord... Why would anyone with sufficient self esteem employ such a trick...
I'm known to many friends as a type that is hard to understand. When I say something, my close friends had to think that I might not talking about what they think I'm talking about...
BUT, using unclear language or grammar or terminologies, are against my rules. Many people intentionally use such "trick" to appear more intelligent. But to me it looks like the opposite 😱
Classic example of criterial biasing. The criteria apparently chosen (is there a difference that I can readily see by some unstated means) and the relevant technical criteria (is there a difference that can be reliably discerned) are not the same criteria.
JCX and I for example appear base our judgements on whether or not there is a difference that can be reliably discerned in z 44 KHz sampled wave which conforms to general technical practice.
An illustration:
Two obviously distinct files with features < 5 uSec apart
The results of downsampling to 44 Khz, showing that while they may look the same, they are in fact still different.
To use Mr. Konchur's astronomy example the last time I looked the impulse response (PSF) of a good even 1/4 mic would preclude resolution of 5us even at 24 bits.
You imho completely missed Kunchur´s point, but nevertheless your screenshots illustrate perfectly what Kunchur meant
Btw, what you´ve tried is (again imho) to use no. 2 and no. 3 from Schopenhauer´s list... 😉
They are? Where.
I do know that he has shown the same result with 2 quite different experiments and that the same result has been had by an independent 3rd party.
Certainly the tests seem to have been done with a lot more rigor than Clark's/yours.
dave
the iatd test is actually the same signals in 2 channels, each presented to R and Left ears respectively, the signals different by (microseconds) delay
so discussing, "drawing" the 2 channels of information by overlapping them in one digital channel is a bit different from the experiment
in fact differencing 2 identical signals separated by a delay is a digital filter - a differentiating/high pass filter
this misses the fact that the 2 ears separately encodes the sound at each ear into pulse trains in the tens of thousands of nerves in the and then collects both R,L auditory nerve bundles in one place and compares them
achieving few microsecond discrimination is amazing - but is explainable by correlation, which is a nonlinear process, correlation seems to be something neural nets do well
getting back to the actual digital audio signals for iatd test conditions of 2 channels it is easy to see sub sample resolution: different by microseconds delayed, nanosecond accuracy after 16/44 round trip
I've done this several times here in LTspice:
http://www.diyaudio.com/forums/everything-else/223748-time-domain-distortion-2.html#post3245634
http://www.diyaudio.com/forums/analogue-source/245555-temporal-resolution-6.html#post3697702
so discussing, "drawing" the 2 channels of information by overlapping them in one digital channel is a bit different from the experiment
in fact differencing 2 identical signals separated by a delay is a digital filter - a differentiating/high pass filter
this misses the fact that the 2 ears separately encodes the sound at each ear into pulse trains in the tens of thousands of nerves in the and then collects both R,L auditory nerve bundles in one place and compares them
achieving few microsecond discrimination is amazing - but is explainable by correlation, which is a nonlinear process, correlation seems to be something neural nets do well
getting back to the actual digital audio signals for iatd test conditions of 2 channels it is easy to see sub sample resolution: different by microseconds delayed, nanosecond accuracy after 16/44 round trip
I've done this several times here in LTspice:
http://www.diyaudio.com/forums/everything-else/223748-time-domain-distortion-2.html#post3245634
http://www.diyaudio.com/forums/analogue-source/245555-temporal-resolution-6.html#post3697702
Last edited:
If you can't follow the link you have, why bother giving you another?
An invalid experiment is invalid even if consistene.
Please explain.