ORIGINAL: jmarkham

i'm not sure what the suggested change in workflow is precisely (based on
this thread?)

firstly, unless i'm planning on mastering the project myself, i never consider
dither. this is relegated to the mastering engineer who will apply both limiting
and bit-reduction. imho, the two must be done together to make the best trade-offs
during this process ... and are best done by the same guy who is providing
consistent levels throughout a project.

secondly, triangular dither is but one of many dithering algorithms ... mixing
to a particular algorithm presumes what the mastering engineer may
choose .. and what noise shaping approaches they may decide upon.

that said, it would be a fine addition to Sonar if the Pow/r dither could be engaged
for realtime monitoring (as it can be with the Waves maximizer series). Pow/R is
often the right algorithm .. but not always ... and simulating what a 16bit bus would
sound like would be great.

jeff

In SoundForge (or similar), generate a sine wave at 4Khz. Open the Spectrum analysis and set buckets to 65,536 (top left). You will see that SF doesn't generate a pure sine wave but anyway, now reduce the bit depth to 8 bits with no dither. Refresh the Freqency analysis view. You will see that there is distortion at many different frequencies. Especially above 4 Khz. (You can also hear this distortion)

Can you perhaps see the magnitude of heavy misunderstanding and incorrectness that has followed the original post?

ORIGINAL: no criminal intent

This is a really informative thread that has cleared up alot of things. Just one small question remains however. Should I leave dither on or not.

zumba

ORIGINAL: Rednroll

Undertow, you are confusing many different items in basic digital audio building blocks.

Undertow, do that same experiment in Adobe Audition, and what you will notice is that you no longer "see" these distortions when opening the exact same waveform.

Why is that? I'll tell you why, because I worked very closely with the Sound Forge developers and pointed this out to them about their waveform display at higher frequencies. It's because the SF waveforms don't interpolate an "anti-imaging" filter being applied in the DACs of your sound card, where Adobe audition does. Thus in Sound Forge what you see is not always what you hear. I can prove this to you by capturing a high frequency sinewave with a digital o'scope connected to the output of my sound card and compare that to the waveform in sound forge if you don't believe that. I have an email somewhere where the developer of SF said, "yeah we skimp on the interpolation of our waveforms." That aside, what you are really talking about has to do with the "anti-imaging" filter being applied. It has nothto do with dither being applied or not.

Pattor and undertow, do a google search on "Digital Anti-imaging filters" and do a little reading. You guys are blowing hot air out your arses. While you're out there you might want to do a little reading on "dither" also and get a better understanding of how it works. Just because you two collectively agree you are correct with your information, it doesn't mean you are right. In my opinion you both are collectively wrong.

ORIGINAL: cGar

Two options:

Since your are exporting to Soundforge - export at 32-bit or 64-bit float. This avoid everything that is depated in this thread. No truncation, no dither, no worries!!!

If you must export at 24-bit: then dither with triangular dither as this is better than the truncation distortion you WILL incur.

ORIGINAL: SteveJL


Dither is NOT required for monitoring, as the output file is 24-bit and it goes through 24-bit DACS, PERIOD.

Dither is ONLY required during the FINAL Mastering stage.

Dither should ONLY be applied ONCE through the whole process, as it is distortion, and is cumulative.

Final PERIOD <g>.

Rednroll ---
I'm gonna read back up on DACs and what not but I was of the impression that once you truncate without dither, that harmonic distortion is there to stay!!! If you are correct and the DACs remove this distortion then you woul also be correct in saying that its better to have dither off. But I do not think you are correct in that. Anti-alias filters stop new harmonic distortion from being added during the DtoA process. The truncation distortion is already present.

ORIGINAL: Rednroll

Ok, I will admit some of my information is not exactly 100% accurate because I did not go into a lot of the details.

The problem that I'm sticking too is that you are applying dither in the mixing process. Well when you apply dither in the mixing process, this does reduce some of the harmonic distortions, but it also increases the overall noise floor.

Later you will be doing a mastering process where that noise floor will be increased in level due to additional compression, and then you will be adding additional noise when you add dither in that process also.

So as I originally said, it is best not to dither and to export the rendered .wav file at the same bit debt as your DAC is capable of, so therefore you are hearing exactly what is coming out of the DAC without any dither applied.

Then when you master the audio and go from 24 bit to 16 bit, I would apply the dither in this process, because now the harmonic distortions are entering into levels that can be heard and you are not adding twice the dither noise to the audio signal.

If your MIX is the final step, prior to your final destination then by all means apply dither, but if it is not then turn it off. As I outlined, the mix is usually not the final step, and I still stick by dither should be used once and only once in the process before going to the final destination.

The problem with the Nika Aldrich information you are referring too (yes, I have read this) is that the information does not address what the DAC does to elliminate these harmonics. His spectral curves are taken Pre DAC output. DACs use "Anti-imaging" and "Interpolation" algorithms to remove those harmonic frequencies due to the square wave stepping. That is what I am getting at.

His study focuses in on the "imaging" effects due to quantization levels in bit reduction. This is true of what happens, but it ignores what a DAC does in the D/A conversion to elliminate these problems.

Why does one 24 bit DtoA converter sound better than another? It's because of their interpolation algorithms, and anti imaging filter sections.

None of the information in the Nika Aldrich report addresses these sections within a DAC, but it does point out the higher noise floor due to adding dither at the cost of reducing the harmonic distortions due to the stair stepping effect. Show me some spectral graphs that have gone thru a D/A converter after they have gone thru the interpolation over sampling algorithms and the anti-imaging filter.

This is what's wrong with your information, because you're only using 1/2 of what is happening and it's ignoring all the processes built into a DAC to overcome these same things that Nika Aldrich outlines for the benefits of adding dither.