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: Dynamic range in high fidelity systems (Read 52203 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Dynamic range in high fidelity systems

Reply #75 –
@Gregorio:

So what you are saying is that if you redfine noise as signal then in order to preserve it accurately then it needs to be dithered?

OK, I can go along with that. 

The only point that I was ever trying to make is that dither doesn't necessarily increase dynamic range, i.e. if the signal noise is large compared to the digitization noise then dither will not make it any quieter, just make it into better-sounding noise.

Dynamic range in high fidelity systems

Reply #76 –
Quote from: Gregorio on
Quoted from the article I posted a link to above: "The bottom line is that WHENEVER a signal goes from a higher resolution to a lower resolution it is necessary to dither in order to avoid the artifacts provided by truncation that have been shown above. This means that whenever signals go from 48 bit resolution for processing to 24 bit resolution, or 24 bit resolution for mixing to 16 bit, or even analog (infinite resolution) to 24 bit during A/D conversion dithering needs to happen."


By what definition of the word does analog have 'infinite' resolution?

Dynamic range in high fidelity systems

Reply #77 –
Quote from: Arnold B. Krueger on
Since virtually every good audio ADC is sigma-delta, they all benefit from the inherent randomization of quantization error and thus as you said, need no additional dither.


Um, delta-sigma convertors do not randomize the quantization error, they noise-shape it to be well above the audio passband.

Just FYI.

However, in order to assure no whistles, birdies, etc, you STILL should be dithering the delta-sigma convertor, but it's a different process to say the least, you might check out Lipshitz/Vanderkoey's papers, which are a bit overstated, but ok from the math point of view.
-----
J. D. (jj) Johnston

Dynamic range in high fidelity systems

Reply #78 –
Quote from: pdq on
@Gregorio:

So what you are saying is that if you redfine noise as signal then in order to preserve it accurately then it needs to be dithered?

OK, I can go along with that. 

The only point that I was ever trying to make is that dither doesn't necessarily increase dynamic range, i.e. if the signal noise is large compared to the digitization noise then dither will not make it any quieter, just make it into better-sounding noise.


Dithering doesn't make quantization error any quieter, it just makes it less noticable. The problems arise when quantization error is correlated with the signal and large enough to be heard. Ever hear undithered quantization erorr? It is the darnedst thng!

Adding more dither doesn't generally make properly-dithered noise sound any different.  Once quantization error is decorrelated from the signal so that any corelation would be vanishingly small, it is hard to decorrelate any futher in a useful way. Since adding more dither adds a tiny amount of noise, you do slightly raise the noise floor. 

Dynamic range in high fidelity systems

Reply #79 –
Quote from: Gregorio on
Quoted from the article I posted a link to above: "The bottom line is that WHENEVER a signal goes from a higher resolution to a lower resolution it is necessary to dither in order to avoid the artifacts provided by truncation that have been shown above. This means that whenever signals go from 48 bit resolution for processing to 24 bit resolution, or 24 bit resolution for mixing to 16 bit, or even analog (infinite resolution) to 24 bit during A/D conversion dithering needs to happen."

Analog recordings are not true analog in that they do not have infinite time and infinite dynamic range.

Paul

     
"Reality is merely an illusion, albeit a very persistent one." Albert Einstein

Dynamic range in high fidelity systems

Reply #80 –
Quote from: Paulhoff on
Quote from: Gregorio on
Quoted from the article I posted a link to above: "The bottom line is that WHENEVER a signal goes from a higher resolution to a lower resolution it is necessary to dither in order to avoid the artifacts provided by truncation that have been shown above. This means that whenever signals go from 48 bit resolution for processing to 24 bit resolution, or 24 bit resolution for mixing to 16 bit, or even analog (infinite resolution) to 24 bit during A/D conversion dithering needs to happen."

Analog recordings are not true analog in that they do not have infinite time and infinite dynamic range.

Paul

     


What many people don't realize is that *nothing* in real world analog has infinite dynamic range. It is another one of those things that seems intuitively clear to some, but is completely wrong. One of the reasons that the mainstream audio world embraced digital so rapidly is that it lacks the relatively high noise floor of real-world analog.

Dynamic range in high fidelity systems

Reply #81 –
Quote from: greynol on
Realizing I sound like a pedant, you originally said "necessary" not "better".  As far as this forum is concerned, whether it is better (let alone necessary!) in any given instance must be determined by a double blind test and not by graphs.

I'm coming more from a professional user's point of rather than from a researcher's point of view. Professionally, it would of course be completely impractical to ABX every track while mastering. Enharmonic quantisation distortion is, AFAIK, a widely accepted phenomena by both the professional and scientific communities. If you need to ABX it for your own piece of mind, it shouldn't be too difficult for you to set one up. For myself, I accept the science and, after experienced quantisation distortion first hand, came to the conclusion as early as 1992 that I would rather pay the price of a higher noise floor using even non-noise shaped dither than to run the risk of quantisation distortion. With modern noise-shaping dither algorithms the perceptual noise floor is so low that the dynamic range is actually extended beyond the theoretical limit of the 16bit file format, so the decision to use noise-shaped dither is a real no-brainer!

Quote from: pdq on
@Gregorio:

So what you are saying is that if you redfine noise as signal then in order to preserve it accurately then it needs to be dithered?

OK, I can go along with that. 

The only point that I was ever trying to make is that dither doesn't necessarily increase dynamic range, i.e. if the signal noise is large compared to the digitization noise then dither will not make it any quieter, just make it into better-sounding noise.

I'm not redefining anything that has not been redefined before. There's white noise, pink noise, brown noise, etc., and then there's the real world noises like wind noise, Brownian motion, electrical interference, room tone, etc. These real world noises are all still defined as noise but are they all statistically random enough to cause the dither effect or are they going to correlate with the signal and cause quantisation distortion? While we are at it, let's be clear (if I haven't been already), quantisation distortion is not just "worse sounding noise", it can also be enharmonic tones which are very obvious and disconcerting. Please, have another look at the article by Nika Aldrich, specifically at figure 4. Truncation hasn't caused the the 100Hz sine wave to have more random noise, it's caused it to have a whole set of enharmonic distortions which would result in the sine wave not sounding like a pure 100Hz sine wave anymore. Dither is applied not to preserve the tonal quality of any noise in the signal but to preserve the tonal quality of the entire signal which, in the case of figure 4 of the article, is not just the noise but more importantly the 100Hz sine wave itself. Taking it as read that we want to maintain the fidelity of a signal we must therefore apply dither, which in effect converts these enharmonic distortions into random noise. Noise-shaped dither goes one step further by making this resultant noise virtually imperceptible, thereby increasing the dynamic range. ... Have you got it now?

Quote from: Paulhoff on
Analog recordings are not true analog in that they do not have infinite time and infinite dynamic range.

What? Do you know of any musicians who have infinite time and dynamic range? Also, if it took an infinite amount of time to make an analog recording, when would anyone have the time (or desire) to listen to it? If it takes infinite time to record and infinite time to listen to, you would therefore need at least infinite time x 2 (plus some extra time for the odd coffee break)? Actually you would need infinite time x 3, because presumably it would take an infinite time to rewind the tape! 

Quote from: krabapple on
By what definition of the word does analog have 'infinite' resolution?

What I took Nika to mean by this statement is that an analog sine wave has a constantly varying voltage. For example, if we take two measurements of the amplitude of the first rising portion of a sine wave, the two measurements will never have quite the same value no matter how small the time difference between the measurements. As we get nearer to an infinitely small time difference between the measurements, we also get nearer to an infinitely small difference between the values of the measurements but there is always a difference (however small). This is in contrast to digital audio which has finite resolution because if the two measurements fall within a single sample period then the values of the two measurements will always be identical. In this context, one could say for all practical purposes that analog has infinite resolution compared to digital.

Cheers, G

Dynamic range in high fidelity systems

Reply #82 –
@Gregorio,

go and look at what lossyWAV does, and then come back and tell me dither is always necessary at 24-bits, or even 16-bits


This isn't rigorous (If fact it's not even valid or true), but as a though experiment, try to imagine that the existing environmental and electronic noise reaching a 24-bit converter without dither can be restated as two noise sources: correct 1-bit RMS TPDF dither at the 24th bit level, plus whatever else is left (i.e. the original noise, minus 1-bit RMS TPDF dither).

Now try to imagine any real world signal - anything you can connect to a 24-bit DAC - where there's any meaningful difference between the original noise, and the original noise minus correct dither.

My conclusion is that the original noise can already be thought of as including dither.


I can easily think of signals which break this for 16-bits, but not for 24-bits. If you can, where are you getting them from?

Remember that TPDF dither only decorrelates the first two moments of the quantisation error* - it leaves higher order moments of error intact (though they're inaudible as far as anyone has yet proven). It's likely that the huge level of noise in any signal wrt the LSB of 24-bit audio already decorrelates better than 1-LSB RMS TPDF dither. So how can you say that it's essential.


I buy the theoretical argument as to why it's a good idea. I just defy anyone to find any case in practice where it makes any audible or measurable difference. I mean with real world signals - anyone of use can generate synthetic signals to prove the point.

Cheers,
David.
P.S. - * - I don't claim to understand this - it's in lipschitz and vanderkooy's classic paper(s).
P.P.S. (seeing your later posts) - IIRC Nika Aldrich didn't really understand dither properly until certain people here explained it properly to him, so tread with care.

Dynamic range in high fidelity systems

Reply #83 –
Quote from: Gregorio on
Quote from: Paulhoff on
Analog recordings are not true analog in that they do not have infinite time and infinite dynamic range.
........
What? Do you know of any musicians who have infinite time and dynamic range? Also, if it took an infinite amount of time to make an analog recording, when would anyone have the time (or desire) to listen to it? If it takes infinite time to record and infinite time to listen to, you would therefore need at least infinite time x 2 (plus some extra time for the odd coffee break)? Actually you would need infinite time x 3, because presumably it would take an infinite time to rewind the tape! .....

You miss the point, I guess. But I have heard from my friend many times say that the vinyl records he loves, or used to love is analog and that there is more on it than a CD can hold. Thanks to woodinville (JJ) I have an good and true answer for that misunderstanding of his. Vinyl has limits, it does not have infinite dynamic range and does not have infinite time or an other way of saying it, infinite frequencies. Tape, Vinyl or whatever is limited and can be quantified.

Paul

     
"Reality is merely an illusion, albeit a very persistent one." Albert Einstein
 

Dynamic range in high fidelity systems

Reply #84 –
Quote from: Paulhoff on
You miss the point, I guess. But I have heard from my friend many times say that the vinyl records he loves, or used to love is analog and that there is more on it than a CD can hold. Thanks to woodinville (JJ) I have an good and true answer for that misunderstanding of his. Vinyl has limits, it does not have infinite dynamic range and does not have infinite time or an other way of saying it, infinite frequencies. Tape, Vinyl or whatever is limited and can be quantified.


I guess maybe you have missed the point. Not only does tape, vinyl, digital and all recording equipment of any type always have limits, so does anything which you would want to record. No sound which has ever existed has infinite frequencies or infinite amplitude.

Quote from: 2Bdecided on
@Gregorio,

I buy the theoretical argument as to why it's a good idea. I just defy anyone to find any case in practice where it makes any audible or measurable difference. I mean with real world signals - anyone of use can generate synthetic signals to prove the point.

Cheers,
David.
P.S. - * - I don't claim to understand this - it's in lipschitz and vanderkooy's classic paper(s).
P.P.S. (seeing your later posts) - IIRC Nika Aldrich didn't really understand dither properly until certain people here explained it properly to him, so tread with care.


Actually, I'm coming from a practical perspective (as an audio professional) and freely admit to not understanding all the math and science behind the processes involved with digital audio theory. However, I have clearly heard the effects of enharmonic quantisation distortion on a real world music mix, in fact even before I knew such a thing as enharmonic quantisation distortion existed! Of course, my experience is, as far as others are concerned, is just anecdotal, so have a look at these two threads from the ProTools forum in 1999:

http://duc.digidesign.com/showthread.php?t...+dithered+mixer
http://duc.digidesign.com/showthread.php?t...+dithered+mixer

In short, some people were noticing artefacts in the signal processing when the software truncated 48bit signals to 24bit signals. DigiDesign themselves took the situation seriously as a considerable number of high profile professionals were involved. After exhaustive tests and discussion (using real music), DigiDesign eventually created a plug in to add dither to the 48bit->24bit reduction process. Hopefully this will provide you with a practical real world case where it made both an audible and measurable difference and of course this is with 24bit resolution, not 16bit!

Reading a little from wiki about lossywav; it implies to me that some sort of dithering process is taking place: "Bit removal adds white noise to the output, however the level of the added noise associated with the removal of a number of bits has been pre-calculated ..."

Cheers, G

PS. I'm sure there was a time when Nika didn't understand the dither process very well, but that doesn't mean to say that after some instruction, learning and research he can't develop into an expert. He may not be the world's greatest expert but he is very highly respected in the professional audio community and I don't feel any particular trepidation about quoting him or a number of other experts.

Dynamic range in high fidelity systems

Reply #85 –
Quote from: Gregorio on
Reading a little from wiki about lossywav; it implies to me that some sort of dithering process is taking place
Dither was available, but never defaulted. It may even have been removed completely now. lossyWAV chops off (truncates) the last X bits (calculating X to be inaudible). Noise shaping is optional (also not defaulted in most versions).

I'll read those long digidesign threads when I get chance.

Cheers,
David.

Dynamic range in high fidelity systems

Reply #86 –
Quote from: Gregorio on
Quote from: krabapple on
By what definition of the word does analog have 'infinite' resolution?

What I took Nika to mean by this statement is that an analog sine wave has a constantly varying voltage. For example, if we take two measurements of the amplitude of the first rising portion of a sine wave, the two measurements will never have quite the same value no matter how small the time difference between the measurements. As we get nearer to an infinitely small time difference between the measurements, we also get nearer to an infinitely small difference between the values of the measurements but there is always a difference (however small). This is in contrast to digital audio which has finite resolution because if the two measurements fall within a single sample period then the values of the two measurements will always be identical. In this context, one could say for all practical purposes that analog has infinite resolution compared to digital.

Cheers, G



My point was that no real-world analog recording has 'infinite resolution'...certainly not in the sense that 'audiophiles' (typically LP philes) claim it does. 


If it's simply continuously varying voltages you're referring to, the sine wav that's output from a DAC after reconstruction also varies continuously, like the one that was input into the ADC.  IOW, to state something I'm sure you know, I *know* Nika knows,  and I which has been stated numerous times on HA, the output of a digital playback system is not a 'stairstepped' wave..it's a smoothly varying sine wav.

Dynamic range in high fidelity systems

Reply #87 –
Quote from: krabapple on
My point was that no real-world analog recording has 'infinite resolution'...certainly not in the sense that 'audiophiles' (typically LP philes) claim it does. 

If it's simply continuously varying voltages you're referring to, the sine wav that's output from a DAC after reconstruction also varies continuously, like the one that was input into the ADC.  IOW, to state something I'm sure you know, I *know* Nika knows,  and I which has been stated numerous times on HA, the output of a digital playback system is not a 'stairstepped' wave..it's a smoothly varying sine wav.


Although I would probably consider myself an audiophile in the true sense of the word, I would not like to be associated with the audiophile movement which seems to me at times to be more closely related to some kind of religious zealotry.

Not being a mathematician, I've always found the term "infinite" to be quite confusing. For example, a sub-set of infinite can also be infinite, and this is what I took Nika to mean. I take it as read that no sound can have an infinite dynamic range. I am aware that the reconstructed signal from a DAC is not the stepped representation of the digital data stored on a hard disk but a continuously varying signal which, at least in theory, is identical to the analog signal before the original quantisation. In other words, although the digital data itself has finite resolution the reconstructed signal does not. Meaning that the 144dB dynamic range of say 24bit when reconstructed could be divided infinitely!  I think you know what I mean and that I am in total agreement with you!

Cheers, G

Dynamic range in high fidelity systems

Reply #88 –
Quote from: Gregorio on
I'm coming more from a professional user's point of rather than from a researcher's point of view.

I'm coming from this forum's point of view.  We are not necessarily only audio professionals, researchers, scientists, engineers, hobbyists or audiophiles.  When you registered you agreed to abide by our rules (TOS #8 in particular).  This does not mean that you get to dismiss them because you're only speaking to a particular crowd.

There are people who come here thinking that 24bits are necessary for a more life-like representation of vinyl upon playback.  Needless to say there is ample reason to suspect that this is overkill.  My suggestion here is that 16 bits truncated is still going to be overkill and it seems that you disagree.  I respect your knowledge on the subject and don't see any reason not to dither since it's free and I don't see that it would cause any harm.  However, I take issue with the idea that it is necessary and challenge you to give me just one example demonstrating this.

Dynamic range in high fidelity systems

Reply #89 –
Quote from: greynol on
Quote from: Gregorio on
I'm coming more from a professional user's point of rather than from a researcher's point of view.

I'm coming from this forum's point of view.  We are not necessarily only audio professionals, researchers, scientists, engineers, hobbyists or audiophiles.  When you registered you agreed to abide by our rules (TOS #8 in particular).  This does not mean that you get to dismiss them because you're only speaking to a particular crowd.

There are people who come here thinking that 24bits are necessary for a more life-like representation of vinyl upon playback.  Needless to say there is ample reason to suspect that this is overkill.  My suggestion here is that 16 bits truncated is still going to be overkill and it seems that you disagree.  I respect your knowledge on the subject and don't see any reason not to dither since it's free and I don't see that it would cause any harm.  However, I take issue with the idea that it is necessary and challenge you to give me just one example demonstrating this.


First of all, I was not denigrating or dismissing anyone. I was simply pointing out that ABX'ing recorded tracks is impractical in a professional's situation (say for a mastering engineer) but not say in a researcher's situation. I am not dismissing researchers at all, simply pointing out that my background is different, not better or worse, just different. Sounds like you may have a bit of a chip on your shoulder? Secondly, I have already provided you with an example proving enharmonic quantisation distortion in Nika's article (figures 3 and 4). If you are talking about an actual audio example, I don't have the facility at the moment to provide you with one but it should not be particularly difficult for you to repeat Nika's experiment. Lastly, my position is as follows:

1. Dithering may not make an audible difference (improvement) under every circumstance.
2. It is impossible to predict when correlation is going to occur and cause enharmonic quantisation errors.
3. There is not the time or facility in the workflow of most professional audio engineering situations to carry out controlled ABX sessions between truncated and dithered mixes.
4. Noise-shaped dither is easy to use, takes virtually no time, will eliminate all traces of enharmonic distortion and will not damage the fidelity of the mix, even if there would have been no enharmonic distortion from truncation.

Baring all these points in mind, in a commercial audio engineering situation, there is no alternative to applying noise-shaped dither and therefore it is a necessity. Although I appreciate this may not be true in a scientific or research situation, the point I was trying to make when differentiating my professional background from the background of a researcher.

Cheers, G

Dynamic range in high fidelity systems

Reply #90 –
Quote from: Gregorio on
First of all, I was not denigrating or dismissing anyone.
I meant dismissing the rules of this forum.  If you wish to make claims that a particular process is going to make an audible difference it is up to you to demonstrate this difference by way of a double blind test.

I can easily provide a vinyl clip digitized to 24 bits and a 16-bit truncated version and an ABX log indicating that I could not reliably demonstrate an audible difference, but it proves nothing.

Dynamic range in high fidelity systems

Reply #91 –
Is it really necessary to continue fighting. From an outside perspective I can tell that you all seem have thorough knowledge of the matter. And from that I conclude, that you probably all understand exactly from what direction each other is coming from and for what or why he might still be fighting. Let's take each other by the hand in a circle, close our eyes, breathe the stubbornness away, and then all smile at each other!

PS The icon palette is seriously missing some peace and love inspired symbols... 

Dynamic range in high fidelity systems

Reply #92 –
Quote from: Gregorio on
First of all, I was not denigrating or dismissing anyone.


Sure you were. You were unrepentantly breaking the TOS you agreed to. If that isn't dismissive and disrespectful of this forum, its participants, its management and even your own integrity, I don't know what is.

Quote
I was simply pointing out that ABX'ing recorded tracks is impractical in a professional's situation (say for a mastering engineer) but not say in a researcher's situation.


This is not true. In the early days of the development of ABX, we did  ABX tests in working, high end recording studios. The results of one of them was published in the JAES. 

This was in the days before digtial/computer audio became common, and where appropriate, involved dedicated ABX boxes that were rare and costly.

These days virtually everything that is needed to do ABX tests exists in the digital domain as a matter of course. 24 bit recordings exist as a matter of course.  Software ABX boxes are freely downloadable. High performance audio interfaces and monitoring systems attached to computers are commonplace.  If *any* professional audio production person really wants to put his beliefs to an ABX test, he can do it, and quite easily, quickly and at no out-of-pocket cost. None!

I'm not suggesting that people use ABX in the middle of a tracking, mixing or mastering session.

But, any person who alleges that their interest in audio is at the professional level can resolve questions like 16 bits versus 24 bits with maybe an hour of easy work. The fact that people aren't doing this is either IMO ignorance or arrogance.

Dynamic range in high fidelity systems

Reply #93 –
Quote from: Gregorio on
Lastly, my position is as follows:
1. Dithering may not make an audible difference (improvement) under every circumstance.


To say the least!  You've been challenged to demonstrate this with a reliable listening test using your own files or files provided to you, and so far you've sloughed the opportunites in violation of TOS 8.

Quote
2. It is impossible to predict when correlation is going to occur and cause enharmonic quantisation errors.


Correlation is absolutely predictable if have the relevant digital data files. There is nothing unpredictable about quantization distoriton in digital systems because all the variables are inhrently precisely quantized and therefore known.

Quote
3. There is not the time or facility in the workflow of most professional audio engineering situations to carry out controlled ABX sessions between truncated and dithered mixes.


That is not what you are being asked to do. There are always various times when people can step outside of the normal audio production workflow and do relevant or irrelevant technical work. If this were not true, nobody would ever eat or sleep. No equipment would ever get maintained. There would never be any equipment upgrades. There would be no training or other technical development.

Stepping outside the normal audio production workflow to do some ABX tests would logically be considered "professional development" or education. Maintenance for the brain, if you will.


Quote
4. Noise-shaped dither is easy to use, takes virtually no time, will eliminate all traces of enharmonic distortion and will not damage the fidelity of the mix, even if there would have been no enharmonic distortion from truncation.


That is a truism.  There has been no controversy over that for decades. The  controversy at hand relates to whether or not 24 bit coding provides a useful sound quality advantage for distributed media.  There is no signficiant controversy over whether or not there are times that using 24 bit coding during audio production can be advantageous. 


Dynamic range in high fidelity systems

Reply #94 –
Quote from: Gregorio on
In practice, the situation is a little more complex than the basic theory suggests but even more weighted to the argument that for the consumer there is no point in higher than 16bit products.

So, where's the controversy?

Dynamic range in high fidelity systems

Reply #95 –
Quote from: greynol on
Quote from: Gregorio on
First of all, I was not denigrating or dismissing anyone.
I meant dismissing the rules of this forum.  If you wish to make claims that a particular process is going to make an audible difference it is up to you to demonstrate this difference by way of a double blind test.

I can easily provide a vinyl clip digitized to 24 bits and a 16-bit truncated version and an ABX log indicating that I could not reliably demonstrate an audible difference, but it proves nothing.


+1, because it is best if people work this sort of thing out for themselves.  Let them pick the program material, let them pick the listening environment, let them do the listening.

I wonder what the average time spent disagreeing with what is reliably known about 24 versus 16 bits is, as compared to how long it takes to do one's own homework. 

IOW do people usually spend hours, days, weeks, months or years pontificating over this issue before they start actually listening to the music?

In many cases it seems to take some people more than a lifetime.  :-(

Dynamic range in high fidelity systems

Reply #96 –
Quote from: Gregorio on
http://duc.digidesign.com/showthread.php?t...+dithered+mixer
http://duc.digidesign.com/showthread.php?t...+dithered+mixer

Quote from: 2Bdecided on
I'll read those long digidesign threads when I get chance.

OK, I've dipped my toe in there, but I think you're going to have to help me if you're familiar with the discussion - at what point in those nearly 200 posts is the issue pin-pointed and resolved?

The thing is, if you're talking about rounding/dithering/whatever within an IIR filter loop, you could be doing that operation thousands, potentially even millions of times on each single sample of a single track of audio. No one will argue against the tiniest change potentially being audible in this case.

If you're talking about once per effect per track, it's far less important. If you're talking about once per track, it's far less important still. If you're talking about once per mix, it's vanishingly insignificant.

So, which is it that's causing the concern in 1999?

I suppose the other important point is that, in some of those cases, depending on how the levels are set at each stage, you might have nominally 24- or 48-bits of audio, but several of the top ones are empty, meaning you have less than you think. e.g. 20-bits, 18-bits, 16-bits - or maybe less if you're really careless!

Cheers,
David.

Dynamic range in high fidelity systems

Reply #97 –
Quote from: 2Bdecided on
Quote from: Gregorio on
http://duc.digidesign.com/showthread.php?t...+dithered+mixer
http://duc.digidesign.com/showthread.php?t...+dithered+mixer

Quote from: 2Bdecided on
I'll read those long digidesign threads when I get chance.

OK, I've dipped my toe in there, but I think you're going to have to help me if you're familiar with the discussion - at what point in those nearly 200 posts is the issue pin-pointed and resolved?


Start out with post one, thread one:

"He basically said that pro tools resolution is not high enuff once you combine all those tracks. For every track you add, you need another bit. I think that means a 24bit session with 32 tracks would require 56 bits on the path."

It's almost like I heard an echo of that here on HA in the past week. Not exactly the same but I definately heard misapprehensions about large scale summing in the digital domain.

AFAIK in 1999, the Pro Tools product did have some technical difficulties. If memory serves, they went through 3 different word lengths for their accumulators. I don't know where their journey had stopped in 1999, but by today all seems well in that department. I seems to recall recent reports that PT SRC might not be in the same league as Audiition and other better compettiive products. The lesson is that you only get what you pay for if you are careful and lucky! ;-)

Dynamic range in high fidelity systems

Reply #98 –
Quote from: Arnold B. Krueger on
Start out with post one, thread one:

"He basically said that pro tools resolution is not high enuff once you combine all those tracks. For every track you add, you need another bit. I think that means a 24bit session with 32 tracks would require 56 bits on the path."


Maybe I can work for that guy - I'll charge by the hour and with each additional hour... 

Quote from: Arnold B. Krueger on
It's almost like I heard an echo of that here on HA in the past week. Not exactly the same but I definately heard misapprehensions about large scale summing in the digital domain.


Well you know how bad computers are at math... 

drewfx