Pavel,
As you are aware there is a slight bit of difference (
) between the theory and the practice. Fourier transforms and Discrete Fourier transforms have differences. So do real A/D's and the math often shown about using them. What I think you are approaching is in the early CD work there was a high frequency boost used in recording and the matching cut in the reproduce end. The idea was to reduce HF noise as was done in analog records and tape. Most folks found that it reduced the quality of reproduction in digital even though it helped in analog. One explanation that may have some merit is that when the HF was boosted it then was encoded by more bits and each additional bit of encoding would add it's own errors even if they were below the LSB. So there were way more distortion products.
So without an HF boost even though there was more uncorrelated noise it sounded better than less but more correlated noise. (OPINION)
So there is merit in using math on an oversampled audio stream to increase the frequency dependent resolution by ratiometric means even though it does not offer an improvement at the highest frequencies or to the full bandwidth.
But worthy of note is that those CD's produced using advanced techniques may sound better than ones that are truly 16 bits! They almost certainly will sound better than the old 9 + 7 bit ones! But of course even from day 1 there were advocates of CD's were perfect or some such.
SY,
Those who can do, those who can't..... Thanks for teaching us your definitions!
Scott,
I just looked at the AES website, no real search. Dither was clearly known in the 70's, but the argument it wasn't really picked up until much later, really holds true even today! But there was a similar case with slew rate distortion, well documented before it became an audio issue.
Now if you want to talk about really new stuff.... well actually there ain't much there. So let us continue arguing about this stuff for fun anyways. Every so often someone actually sites a paper in context and we can learn stuff that is new to us!
John,
Forgive me no time to follow the tradition on this thread and insult you, maybe I'll have time later this week!
ES
Those who can do, those who can't..... Thanks for teaching us your definitions!
Hard-won knowledge from doing. We were using dither on digitized signals from mikes back in the late '70s and it wasn't new then. Admittedly, we were picking up acoustic signals from molecules, not guitars.
Scott, you are right. I should NOT overemphasize what CEDAR does to program material. Sometimes it makes it better, overall, BUT sometimes when used indiscriminately, it can remove the ESSENCE of the program material. Yes, the hiss is gone, but so is all the interesting information. We might as well hum the tune to ourselves from memory.
I have one on my groin from that. Seriously.
Speaking of groin busters I just unloaded the speakers. Great guy a little OCD and they are pristine, but woofer was railed positive, quite toast. We said hi shook hands and the first thing he said was "Your wife must be more tolerant than mine", I kid you not. These things have negative WAF.
Should have some dither pictures tonight to talk around, realized truncation of 32bit to 16bit with no dither produces the same data as ideal 16bit A/D so I can do simulations with nothing but Audition. I have some tests that might be interesting.
Unless you're using quite different software that I am, 'asynchronous' is a misnomer. Even 192 kHz to 44.1 kHz is synchronized. It requires a 28.224 MHz intermediate sample rate, but every sample falls precisely on a clock in each step.
There are ASRC hardware chips which reclock during SRC, but those allow noise incursion.
I call shenanigans. You distracted me by mentioning ZOH, which is what was used to create some of the bad HD releases I've witnessed. Your example of "High Rez" to CD is simply not possible with ZOH. There is no time in which to "Hold" a sample when you reduce the rate. Tossing samples is called decimation, not ZOH. You're not just trolling, are you?
When you open them, expect to find some spectacularly blown capacitors.
was Bob Fulton the first person to mention (home hifi) wire quality,
in these united states?
and what of other countrys, certainly other people, in other places,
have made improvements .
(except maybe Nunavut,
where all wire is in a state of constant cryogenisis.)
Hi,
Dither is not gaussian.
The visible result is described very well by this term.
My issue is not +/- 1LSB dither in a 24 Bit ADC or DAC or even in a 16 Bit one. It is the use of "Dither" as the sole or major means to create A2D and D2A conversion.
Well digital recordings as such again are not neccesarily a problem. Not even all professional AD's and DA's use dither.
I am getting more and more interested in recording again and we can now find 16 Bit SAR ADC's (Sucessive Approximation Register Analog to Digital Converters) and multibit DAC's that can handle "mega samples" (in other words 8 times or more oversampling at 192KHz sample rate) and could also be "stacked" if using lower bit types.
Combined with suitable processing this could be used to give around 20 Bit "true multibit" recordings without digital filtering or other processing at sample rates up to 192KHz (lower sample rates allow more resolution). It should be quite interesting. For now anyone can get one of the old PCM system AD/DA systems of e-bay and use them for recording and playback sans oversampling, dither et al... If you restore an example well the results can be instructive.
As for record mastering, it used to use long tape loops for delay until digital delays became common place in the 70's and 80's. Some record cutting houses use tape delay based groove spacing control to this day.
Ciao T
Dither is not gaussian.
The visible result is described very well by this term.
My issue is not +/- 1LSB dither in a 24 Bit ADC or DAC or even in a 16 Bit one. It is the use of "Dither" as the sole or major means to create A2D and D2A conversion.
Well digital recordings as such again are not neccesarily a problem. Not even all professional AD's and DA's use dither.
I am getting more and more interested in recording again and we can now find 16 Bit SAR ADC's (Sucessive Approximation Register Analog to Digital Converters) and multibit DAC's that can handle "mega samples" (in other words 8 times or more oversampling at 192KHz sample rate) and could also be "stacked" if using lower bit types.
Combined with suitable processing this could be used to give around 20 Bit "true multibit" recordings without digital filtering or other processing at sample rates up to 192KHz (lower sample rates allow more resolution). It should be quite interesting. For now anyone can get one of the old PCM system AD/DA systems of e-bay and use them for recording and playback sans oversampling, dither et al... If you restore an example well the results can be instructive.
As for record mastering, it used to use long tape loops for delay until digital delays became common place in the 70's and 80's. Some record cutting houses use tape delay based groove spacing control to this day.
Ciao T
It certainly can be. In fact, any real world "24 bit" A/D will be dithered by its own noise in a pretty much Gaussian way. This is not evil.
I'd get tired of those two dozen records pretty quick. Arf.
Thanks,
Chris
Hi,
So, essentially your software upsamples to 147 times of the source 192KHz rate and then downsamples by 640 times to get to 44.1KHz? I mean it REALLY does that?
And what if your source sample rate has a 50ppm Error?
I know nothing of the software you use (as you have omitted to mention it), but I have not come across the kind of solution you suggest in a situation where "live" signals are processed.
We agree on these ASRC chips then...
Traditionally decimation includes lowpass filtering and processing, what I propose does not and I am unaware of a correct term for it. Maybe "direct decimation" then?
Ciao T
So, essentially your software upsamples to 147 times of the source 192KHz rate and then downsamples by 640 times to get to 44.1KHz? I mean it REALLY does that?
And what if your source sample rate has a 50ppm Error?
I know nothing of the software you use (as you have omitted to mention it), but I have not come across the kind of solution you suggest in a situation where "live" signals are processed.
We agree on these ASRC chips then...
Traditionally decimation includes lowpass filtering and processing, what I propose does not and I am unaware of a correct term for it. Maybe "direct decimation" then?
Ciao T
Hi,
Any real word "24 Bit Audio ADC" will actually be a low or single bit modulator with MOST of the "24 Bits" created by dither and this dither is not Gaussian.
I am not debating if this is evil or not, but if it is an appropriate design approach for high quality audio, which I feel it is not, despite having become common practice.
You may find that there are many more than two dozen...
Ciao T
Any real word "24 Bit Audio ADC" will actually be a low or single bit modulator with MOST of the "24 Bits" created by dither and this dither is not Gaussian.
I am not debating if this is evil or not, but if it is an appropriate design approach for high quality audio, which I feel it is not, despite having become common practice.
You may find that there are many more than two dozen...
Ciao T
Hi,
Will it now? Illegal in what way? Who wrote the law?
And who is going to arrest me for having music with "Illegal Values" in it? The Police or Sting?
I am aware the results will violate certain dearly held precepts, but I could care an iota about them.
So, have you tried it?
Ciao T
Will it now? Illegal in what way? Who wrote the law?
And who is going to arrest me for having music with "Illegal Values" in it? The Police or Sting?
I am aware the results will violate certain dearly held precepts, but I could care an iota about them.
So, have you tried it?
Ciao T
Tomtt, not to ignore you. Yes, you are correct, in the USA, Bob Fulton made some interesting breakthroughs in wires, AND he made GREAT test records using elementary school children. I miss them, as they were destroyed in the firestorm. What a way to show off your hi fi! In Europe, it was someone else, from France.
- Status
- Not open for further replies.