Don't know, why do physicists and EE's use different scaling on FFT's? I have not read it carefully enough to see.
I'm glad you mentioned Paul Frindle because it was my intention to mention his work on the SONY OXF-R3 console (which is considered a state of the art digital console costing over 1 million$).
He has a deep understanding of analog and digital and he did some serious research to correlate audibility of various processes.
He discovered that even dither at -110dBFS could be detected in -70dBFS tape noise recorded digitally if in the randomness of the dither there was a partial statistical correlation. Let's not forget that there is no real random in digital computer world, so you have to know how to design a good random generator using limited computing resources.
chrissugar
Hi,
That is completely besides the point.
So you insist on not showing anything real that relates to how things really work while presenting something that is an idealised version of hos things could be done, but are not. That is your prerogative...
Ciao T
That is completely besides the point.
So you insist on not showing anything real that relates to how things really work while presenting something that is an idealised version of hos things could be done, but are not. That is your prerogative...
Ciao T
Hi,
Is it? Would you say that applying 14 Bits of worth of dither and converting a 24 Bit PCM Signal to 10 Bit with this is a good idea?
Just because you say so does not make it so.
You are aware of Ethan Wienders tests?
Grekim and Ethan test dither, jitter, A/D
Funnily enough, that is not what I hear and what for example Gremkin/Wiener found.
And note, I do take umbrage as such with 2LSB dither applied to the 16 Bit CD Release when coming from a 24 Bit Master (though the actual need for said is debatable), but I do take issue with 16384 LSB Dither applied to a 24 Bit signal...
Ciao T
Is it? Would you say that applying 14 Bits of worth of dither and converting a 24 Bit PCM Signal to 10 Bit with this is a good idea?
Just because you say so does not make it so.
You are aware of Ethan Wienders tests?
Grekim and Ethan test dither, jitter, A/D
Funnily enough, that is not what I hear and what for example Gremkin/Wiener found.
And note, I do take umbrage as such with 2LSB dither applied to the 16 Bit CD Release when coming from a 24 Bit Master (though the actual need for said is debatable), but I do take issue with 16384 LSB Dither applied to a 24 Bit signal...
Ciao T
I missed the return to the condecending tone. What I presented was text book stuff. I would think converters used to master CD's are a bit better than off the shelf and their INL and DNL are pretty close to ideal, in any case things are much better now. If you sent a complete map of the INL and DNL of your particular A/D or D/A it could be simulated, but no I don't have time.
As for the harping on FFT's, ideal TPDF dither produces uniformly distributed white noise as a residual, the spectra are not interesting on any scale. Text book stuff again.
Why am I not surprised? It's either that or someone else paid for it so it's proprietary. Gee when I use "idealized" converters it's not OK.
Attachments
Last edited:
I have that paper. He lists a bunch of things that he claims are audible, but neither references the tests nor (for the ones he did) gives any detail whatever on how the tests were done and what precisely was being tested. It's an entertaining document, much of it is useful, but it's rather short on data. I wouldn't be citing it as evidence that a 10us interchannel time delay is audible on speakers.
I do note with amusement that he says several times that the quality of the equipment used in evaluations isn't very important as long as level and frequency response variations are controlled.
You insist on some claims that have no ground in theory and in practice.
My (and many others) measurements and listening tests on real equipment correlate very well with the theory so there is no gap between the two world.
Also all the measurements and listening tests done by Dan Lavry and Paul Frindle are not some theoretical simulations but results with real converters and processors.
here is a small excerpt from Dan's paper:
"Dithered signals provide a "constant" noise floor, independent from signal and DC offset.
Measuring a triangular dither with a "dB meter" shows a reading of -93 dB for 16 bits. The energy
density (at each frequency) is at about -125 dB. Can you hear under the noise floor? You can hear
30dB below your "meter", all the way down to the noise density in the surrounding frequencies.
Reexamination of the frequency plots shows that you can hear a 16 bits dithered signal down to about
-125 dB under full scale.
Some manufacturers choose to view the special case of undithered signal gating as a the 16 bits
hearing threshold. One should not confuse the "gating threshold" of unditherd system with the noise
density of a dithered one. The "special gating case with 1/2 LSB of DC" occurs at about -96 dB. The
noise density (per frequency) for dithered signals is almost 30 dB lower.
The ability of the dither "to to bring the gated signal back" is shared by all types of dither.
Rectangular, Nyquist and triangular all perform the task within about 3 dB of each other (a range of
about 1/2 a bit). Beware of claims for "a special ability" of a specific type of dither to provide "3 -4
more bits". The 30 dB or so of dynamic range "beyond" the gating threshold is not unique to one type
of dither. It is shared by all types of dither and is not to be confused for additional bits. The proper
criteria for dither quality is its ability to eliminate distortions and noise modulation."
chrissugar
Did they evaluate the noise of the original? Lots of opportunity to obviate dither.
If the final delivery format is 10bit then yes but why we are talking about a limited format.
It is Ethan Winer and yes I'm aware for more than 10 years about all of his demistification. His tests are mostly based on mediocre quality recordings. I'm not surprised he thinks the Creative Sound Blaster is as good as an Apogee converter.
Nobody in the mastering comunity is taking him serious.
I think you should listen to some of the guys who proved their hearing and technical abilities with their work like Keith Johnson, Dan Lavry, Paul Frindle, Bob Katz and many others.
I think you are alone because the whole mastering community (you know the guys who deliver all the audiophile music available) think different related to dither.
Dither is not an option in the real world audio.
chrissugar
First, a 10 uSec interchannel broadspectrum delay is easily audible..trivial. But it can only be discerned using headphones. Attempts to counter the image shift using a pan pot has mixed results, as all the image does not shift exactly the same, it is frequency dependent. Some of the content appears to shift easily, some of the content "locks" off axis, and panning necessary to center the locked content shifts other content too much. To hear this, you must use a mono signal of course. Note: this is my personal observations over about 4 years, and of course requires duplication by others.
A 10 uSec interchannel broadspectrum delay in speakers is NOT audible. Humans will simply adjust to the sweetspot shift.
When a SPECIFIC voice or instrument is shifted 10 uSec interchannel with respect to other content which has not been shifted, humans will perceive the drift of the shifted content RELATIVE TO the unshifted.
Who's kidding who?? This thread is already like that...
Cheers, jn
Read down a way... Looks very trustworthy.
"The hum thing is a little disturbing. The SB was looking for a -10 dBu level (I assume) and the Apogee was looking for a +4 dBu level. We cranked the Apogee input "trim" all the way up to make the signal even with the SB input. In retrospect, I should have adjusted a dip switch for -10 nominal level operation which would have required less cranking up the gain. Other explanations could be the path between 1202 and Apogee. I know I've never made a balanced recording with it and had any audible hum.
By the way, hum or no hum, I don't feel this really hurts any validity of what we did. I don't think 60 cycles and its harmonics will affect at all the way something like a triangle will be recorded. And I doubt there was any significant impact on the guitar track either. Of course we traded off running balanced for doing a split to get two simultaneous recordings, a fact that is very important to the whole experiment. "
OK, so my "non-sequiturs" led me to the same conclusion as yours.
Can you suggest a signal where 10us delay in one channel for headphones is "easily audible.... trivial?" I'd like to try it for myself.
Hi,
Practically all modern ADC and DAC's operate preciely like this, even if the writing on the outside says "23 Bit" or even "32 Bit"
Actually, there are reasons why Keith Johnson uses mostly his own designs, even for converters. As to the rest, I am familiar with them. I said it before and will say it again, just because X or Y says so does not make it true.
Ciao T
Practically all modern ADC and DAC's operate preciely like this, even if the writing on the outside says "23 Bit" or even "32 Bit"
Actually, there are reasons why Keith Johnson uses mostly his own designs, even for converters. As to the rest, I am familiar with them. I said it before and will say it again, just because X or Y says so does not make it true.
Ciao T
Scott,
What you presented had NIL to do with how things are applied in reality. Of course you DO NOT HAVE TO show things that bear any relation to reality.
I would think I personally would look up some of the real stuff and check my assumptions in the harsh light of reality, but that's just me... Others prefer to simply AssUMe...
No need to that. Instead show the Dither that is USUALLY applied and not you arguably clever and "a good idea" subtractive version. No need to include the converters as "non-ideal", it is not required for the demonstration.
A squid eating dough in a polyethylene bag is fast and bulbous. Got me?
Ciao T
What you presented had NIL to do with how things are applied in reality. Of course you DO NOT HAVE TO show things that bear any relation to reality.
I would think I personally would look up some of the real stuff and check my assumptions in the harsh light of reality, but that's just me... Others prefer to simply AssUMe...
No need to that. Instead show the Dither that is USUALLY applied and not you arguably clever and "a good idea" subtractive version. No need to include the converters as "non-ideal", it is not required for the demonstration.
A squid eating dough in a polyethylene bag is fast and bulbous. Got me?
Ciao T
OK, so my "non-sequiturs" led me to the same conclusion as yours.
Can you suggest a signal where 10us delay in one channel for headphones is "easily audible.... trivial?" I'd like to try it for myself.
Wooly Bully by Sam the Sham and the Pharoes(sp)
Eli's Coming Three Dog Knight.
Cheryl Lynn To be Real.
Inna Gadda Da Vida Iron Butterfly
Thompson Twins In the name of love.
CJ and Company Devils Gun..
Anything you want.
You may spend days and days reversing headphones, reversing channels electrically, altering the panning without looking...as long as it takes. I've done it for years. Constantly testing my results every chance I get...
You set the parameters, make yourself happy that you are not just imagining anything..you know, standard protocol..
Cheers, jn
ps. The verbage "non sequitur" was because I specifically stated "with respect to non shifted content", or sumptin like that.
I was talking about classic SAR A/D's and R2R ladder DAC's. I could care less about your beef with bitstream stuff.
What does Captain Beefheart have to do with dither?
Last edited:
Hi,
If we are talking about the chip, I2S which with a BCK of 64fs is in effects a 32 bit format. In terms of a system, USB and PCI based Audio can work at 32 Bit and even 384KHz sample rate if needs must.
Of course, most so-called "32 Bit" Dac's and Adc's are actually 8-12 Bit parts with dither and noiseshaping giving at best around 21 Bits ENOB with A-weighting and looking non too closely, only one current part manages actually 22Bit ENOB as DAC. But it still says "32 Bit" on the package...
Ciao T
If we are talking about the chip, I2S which with a BCK of 64fs is in effects a 32 bit format. In terms of a system, USB and PCI based Audio can work at 32 Bit and even 384KHz sample rate if needs must.
Of course, most so-called "32 Bit" Dac's and Adc's are actually 8-12 Bit parts with dither and noiseshaping giving at best around 21 Bits ENOB with A-weighting and looking non too closely, only one current part manages actually 22Bit ENOB as DAC. But it still says "32 Bit" on the package...
Ciao T
- Status
- Not open for further replies.