Special Dithering / Bit-Resolution Question

[LukeFromBerlin]

OK, this one bugs me since quite a while (a specialist question)

I prefer RECORDING my mixes on another track rather than bouncing it (for several reasons). As I mix for CD, I have to dither my 24bit session down to 16 bit as the final part of the mastering process of course.
I do this by inserting a mastering/dithering plugin (Ozone 4, whose dithering algorhythms and options I prefer to avid's standard) on an aux track and send the result via a bus to the "final" recording audiotrack. So the recorded audiofile should be -AFAIK- a 16 bit file now, as the additional bits should be empty bits or "0" (zero) after the down-dithering. So although still a part of a 24 bit session and "oficially" a 24 bit audiofile, in "reality" there should only be 16 bits containing information in the audiofile now (after passing the dithering plugin)

BUT: after doing some truncating, perhaps a little fade correction at the end (and consolidating the region after that of course) I have to EXPORT the region as an audio (.wav) file to be able to use it ( = the final and mastered version)

ProTools, not knowing that I have turned the 24 bit file into a "in reality" 16 bit file already (with empty bits above bit nr. 16) will ask me upon export what bit format I want. When I enter "16 bit" it will dither AGAIN, perhaps even do some noise shaping).

My question: does this have ANY EFFECT on the audiofile ? In other words, will PT re-arrange all the bits and mess up my dithering I have done before (on the aux track) or will this second dithering upon "exiting" the 24bit session have no effect at all, as the algorhythm is only trying to reach for the bits above 16 - and won't find any information there ?

Are the files EXACTLY the same down to each single bit after the export/dithering (I cannot avoid) process as the have been dithered before or is everything dithered TWICE, thus messing a lot of stuff up inside the audiofile ?

Help from dithering / bit experts would be greatly appreciated. I have no clue how the dithering works, how it turns 24 bits into 16 bits ? (does it try to keep as much information as possible and MOVE bits around, or does it just cut of any bits above bit nr. 16 ?

Thanx !

[Park Seward]

The final version of audio that goes onto a compact disc contains only 16 bits per sample, but throughout the production process a greater number of bits are typically used to represent the sample. In the end, the digital data must be reduced to 16 bits for pressing onto a CD and distributing.
There are multiple ways to do this. One can, for example, simply discard the excess bits – called truncation. One can also round the excess bits to the nearest value. Each of these methods, however, results in predictable and determinable errors in the result. Take, for example, a waveform that consists of the following values:
1 2 3 4 5 6 7 8
If the waveform is reduced by, say, 20% then the following are the new values:
0.8 1.6 2.4 3.2 4.0 4.8 5.6 6.4
If these values are truncated it results in the following data:
0 1 2 3 4 4 5 6
If these values are rounded instead it results in the following data:
1 2 2 3 4 5 6 6
For any original waveform, the process of reducing the waveform amplitude by 20% results in regular errors. Take for example a sine wave that, for some portion, matches the values above. Every time the sine wave's value hit 3.2, the truncated result would be off by 0.2, as in the sample data above. Every time the sine wave's value hit 4.0, there would be no error since the truncated result would be off by 0.0, also shown above. The magnitude of this error changes regularly and repeatedly throughout the sine wave's cycle. It is precisely this error which manifests itself as distortion. What the ear hears as distortion is the additional content at discrete frequencies created by the regular and repeated quantization error.
A plausible solution would be to take the 2 digit number (say, 4.8) and round it one direction or the other. For example, it could be rounded to 5 one time and then 4 the next time. This would make the long-term average 4.5 instead of 4, so that over the long-term the value is closer to its actual value. This, on the other hand, still results in determinable (though more complicated) error. Every other time the value 4.8 comes up the result is an error of 0.2, and the other times it is −0.8. This still results in a repeating, quantifiable error.
Another plausible solution would be to take 4.8 and round it so that the first four times out of five it rounded up to 5, and the fifth time it rounded to 4. This would average out to exactly 4.8 over the long term. Unfortunately, however, it still results in repeatable and determinable errors, and those errors still manifest themselves as distortion to the ear (though oversampling can reduce this).
This leads to the dither solution. Rather than predictably rounding up or down in a repeating pattern, it is possible to round up or down in a random pattern. Dithering is a way to randomly toggle the results between 4 and 5 so that 80% of the time it ended up on 5 then it would average 4.8 over the long run but would have random, non-repeating error in the result.

http://en.wikipedia.org/wiki/Dither

[LukeFromBerlin]

Thankx Park Seward, I had checked German Wiki already of course, but the English one is a little more thorough.

But I am familiar with the dithering basics, even have a clue how the Pow3 (that AFAIK PT uses) dithering/noise shaping works.

My question is a different one, I'll try it again:

When a 24 bit Audio file is DITHERED to 16 bit, but not CONVERTED to a 16 BIT FILE yet (will be converted in another step) what happens to the audio ?

I presume, after the change of the bitrate / dithering process all information will be inside the first 16 bits per sample, and the additional 8 bits -that still "exist"- will be EMPTY (?), "non existent", or containing a "0" ... ?

And how will the algorhythms treat this audiofile when converting this file to a 16 bIT AUDIOFILE after that ?


Meaning:
(A) When bouncing with the dither plug in on the mastertrack insert, DITHERING and CONVERTING both take place in ONE GO, and BEFORE the file is saved to disk.

(B) When dithering through a Plug in inserted somewhere before the recording audio track, DITHERING and bitreduction take place but NOT THE CONVERTING of the FILE. It will still be an official "24 bit audio file" (containing waveforms reduced to 16 bit and dithered to reduce distortion and artifacts)

NOW:
When this file (a 24bit audiofile containing audio reduced to 16 bit) is CONVERTED to a 16 BIT AUDIO FILE file in the next step (while exporting the region from PT, and PT unfortunately CANNOT export with a different bitrate without dithering) are the first 16 bits changed in any way ?

Does PT interpret it as
1) "no need to do anything", as every sample contains only 16 bits with INFORMATION, the others are of no mathematical relevance
or
2) does it interpret all 24 bits to be "used bits" or bits of relevance and will it average them out, meaning the value of "0" or "empty" bits Nr. 17-24 per sample will significantly mess up the values of bits 1-16 per sample ?

I know, this is a tough one, but perhaps someone out there knows

[Rich Breen]

Originally Posted by LukeFromBerlin

...NOW:
When this file (a 24bit audiofile containing audio reduced to 16 bit) is CONVERTED to a 16 BIT AUDIO FILE file in the next step (while exporting the region from PT, and PT unfortunately CANNOT export with a different bitrate without dithering) are the first 16 bits changed in any way ? ...

You're making this way more complicated than it is; On an export to 16bit, Pro Tools adds 16bit-level dither (noise shaped), and then lops off the bottom 8 bits. ProTools doesn't "know" anything about previous processes that may or may have been done to the source 24bit file. If you'd previously added 16bit dither to the (24bit) file, than your final file will be double dithered (noisier than it has to be, though not by much). Yes, it would be nice if one could control dither from the ProTools export clip dialog box or simply truncate (to allow for previouly dithered files), but you can't. If you want to add your own 16bit-level dither and then truncate to 16bits, and you need to do it in ProTools, then you'll need to use the bounce function (this will be faster and easier in PT11). There are other more specialized programs better suited to a CD prep workflow.

best,
rich

[JFreak]

It is actually not that simple.

AFAIK, dithering is not pure math. By definition, the added "dither" is a random addition to the signal, so you cannot just say some percentage change will affect the end product this or that much.

What is important is the principle. What happens when you reduce bit depth, and why. Don't get into math that much, but take some time to think about the principle. Why dither? What happens? What is noise shaping?

If you think this makes zero sense whatsoever, you can always mix to analog and master from that.

[Park Seward]

This may help:

"Dynamic Rounding - Dynamic Rounding is a technique devised by Quantel for truncating the word length of pixels – a process you can't avoid when you are processing images. Rather than simply losing the lower bits, Dynamic Rounding uses their information to control, via a randomiser, the dither of the LSB of the truncated result. This effectively removes any artefacts that would otherwise be visible. Dynamic rounding is non-cumulative on any number of passes and produces statistically correct results. Dynamic rounding eliminates any truncation artefacts."

So if we think about images rather than sound, Dynamic Rounding is an intelligent dither.

[DDDaniel]

Hi, explaining dithering as randomizing the rounding error is a totally correct way of looking at it. (my compliment for the explanation, never seen it so clearly explained) But there is another way of looking at it, and that is from the sound point of view.
And then dither is an added very low level noise, which is visible on audio frequency meter like the free very handy Bluecat Analyzer.
Depending on kind and shape it is more or less visible, but might not be audible.
So from the tech point of view you change the file as soon as you apply dither. If it is audible is a different question. (most prob. answered different by different people

If you would be able to switch off the dither it becomes a more complex question: when you export a 24 bit file to 16 , you would assume that if you would truncate a 24 bit file with 16 bit dithered content would not change, but Avid in it's white paper kind of says that reality is more complex, and that even when you keep the fader at 0dB the file is recalculated because of the summing in the mixer plugin, but for an export you would assume it would bypass the mixer.
So then it comes down to how the software works...

But again, the question is, would you hear it?

Hope this helps...