In your opinion what is the best setting for dithering? Why?

Great question RFC. I've often wondered this myself. Can't find much documentation on the varies types as well. In my list theres: Rectangular, Triangular and Pow-r1, 2, 3! I've always left it as is on Rectangular, but have often wondered if the other settings are better. Anyone that could shed some light on this subject, we would be surely grateful!

Dithering basically amounts to inserting small amounts of noise in order to improve the psycho-acoustic effect after reducing bit rate.

Rectangular, Triangular, and Pow-r put different amounts of noise into different frequencies. I think the main difference among the Pow-r settings is how much CPU is used; I can't see any reason not to use Pow-r3 -- if I've spent days working on a project, I'm happy to let the CPU work a little harder on the dithering!

I have generally found Pow-r3 to give the cleanest result, meaning that my recordings sound pretty much indistinguishable to me after applying it. By contrast, Rec and Tri both seem, to my ears, to leave the recording sounding just a tad bit duller.

It's easy to experiment. What I did was to take a small 24-bit audio clip, about 20 seconds, exported it using the different dither settings, and compared the results.

i don't use any dithering within sonar. if the bit rate isn't changing, there's no reason to, as far as i know. when converting for export, i use pow-r 3.

ORIGINAL: ru

i don't use any dithering within sonar. if the bit rate isn't changing, there's no reason to, as far as i know. when converting for export, i use pow-r 3.

+1


Wondering if I have misunderstood something?? Isnt dither the final step in the process before mixing down to 44/16?

Interesting Don.

My main question about dithering is when and how to use it?

It's my understanding that dithering adds to the CPU and Power-3 has the biggest hit. So how does dithering add to the CPU, is it while your working on and playing the project? It shouldn't matter during an export.

I'm still useing Sonar-4 (S4) and you have to choose which bit depth to use for each project. Is this still true for S7?

So in S4 if the bit depth has to be one or the other, why would you use dithering while your working on and building the project? Unless maybe your project was setup for 16-bit and you imported any 24-bit files. Heh heh, I've never done this so I'm not sure it can even be done?

Another scenario would be if you have a 16-bit project but you're useing a sampler that has 24-bit samples. I have several projects like that and they seem to work fine whether I've got dithering selected or not. I actually do have the Power-3 dithering selected for most of these projects but I don't have a clue if it's doing anything.

Since the manual says that although Power-3 has the piggest CPU hit, it's also the best, especially for orchestral music. So that's the one I use all the time.

Heh heh, I could be all wet here but if so, I wouldl gladdly and appreciate being enlightened.

T.S.

If you search for dither posts with me as the author, you'll turn up some interesting discussions (not by me in every case!) of the selection of different manufacturer's dither algorithms and applications. Most recently there is a fairly hot debate going on with very polarized sides regarding using and not using dither - pretty funny. About a month and a half ago I chose to stand some ground on the no-dither side just to see what happened - eventually I even brought Bob Katz into the discussion via an email exchange I had just completed with him.

Summary - exciting the LSB of a digital audio file when performing WLR is mathematically appropriate. Your selection of algorithm is dependent upon the material and where the noise shape affects harmonic distortion and linearity/clarity of the of the resulting bandshape.

I continue to use POW-R 3 and Mbit - I find them interchangeable in the delicate sacred choral ensembles I record - the detail of the room decay in some quiet chapels is one of my tests for dither selection. Another is the potential artifacts in upper harmonic distortion but I've not been looking there for some time - there are some fairly well documented reviews of that outcome of WRL and dither.

-D

ORIGINAL: D K

Isnt dither the final step in the process before mixing down to 44/16?

Yes, but there's also bit reduction applied to files when boucing/freezing. For example, if you have bounce set to 32 bit under Options::Global::Audio Data then, since there's a bit reduction involved, Sonar will use the default dithering option you've set under Options::Audio::Advanced during the processing of those files. The exception would be if you had bounce set to 64 bit then Sonar won't apply any dithering of course.

You should never dither, "never".

ORIGINAL: Jose7822

ORIGINAL: jamjar


Even if you are working in a different bit depth to the final bit depth, then you can still turn off dithering until the very FINAL step (ie the final mix). By doing so, you'll might save enough CPU power for that extra reverb you couldn't run before.............. Of course, if you want to hear exactly what the final mix will sound like BEFORE you actually make it, then you'll have to turn it back on.

This is not true. Dithering is ONLY applied during a bounce, freeze or export operation when a bit depth (wordlength) reduction takes place. Sonar DOES NOT dither during playback or when the wordlength are the same. You can hear the final results with dithering while exporting using audible bounce though.


Take care!

Yeah, I guessed that would be the case but had suggested turning it off just in case.

I was once a software engineer and I know enough about the way that world works to know that you should never assume the obvious. We would naturally assume that sonar would bypass certain processing tasks to save CPU when not required, but you can never guarantee that the SW engineers got round to implementing the functionality.

Stuart.

ORIGINAL: jamjar


Yeah, I guessed that would be the case but had suggested turning it off just in case.

I was once a software engineer and I know enough about the way that world works to know that you should never assume the obvious. We would naturally assume that sonar would bypass certain processing tasks to save CPU when not required, but you can never guarantee that the SW engineers got round to implementing the functionality.

Stuart.

I see what you're saying, but consider that Sonar has a 64 bit engine and having dithering ON all the time would mean that you're not hearing a 64 bit engine anymore. You would immediately know it. The bakers are not that bad .



EDIT: If you want to prove it to yourself then insert Cakewalk's bitmeter in the FX Bin of the Master Bus. If all your plugs are 64 bit native (like the Sonitus plugins) then you will see it light up all the way down to 64 bits. Set it to Float though.

ORIGINAL: John

ORIGINAL: Jose7822

ORIGINAL: robby

You should never dither, "never".

Then your music is suffering from the effects of truncation. Even though dither applies noise to the music, which sounds counter-intuitive, it serves the purpose of smoothing out the truncation that happens when reducing wordlengths and will actually make your music sound better. As DonM pointed out, this noise excites the LSB (Least Significant Bit) producing the effect of having a bit depth higher than 16 bit (close to 19 bits!). The trade off is the noise being introduced but, with algorithms like POW-R and MBIT+, this noise is pushed away from the hearing range. Here's more info on the POW-R dithering found in Sonar and other high-end equipment:

http://www.weiss.ch/pow-r/pow-r.html

Remember it's Robby. I think he is using it in its other meaning. As in don't dither around. Or he is in a dither. It was not always used with digital application.

This is true, I totally forgot . Robby, you fooled me .

ORIGINAL: Jose7822

ORIGINAL: jamjar


Even if you are working in a different bit depth to the final bit depth, then you can still turn off dithering until the very FINAL step (ie the final mix). By doing so, you'll might save enough CPU power for that extra reverb you couldn't run before.............. Of course, if you want to hear exactly what the final mix will sound like BEFORE you actually make it, then you'll have to turn it back on.

This is not true. Dithering is ONLY applied during a bounce, freeze or export operation when a bit depth (wordlength) reduction takes place. Sonar DOES NOT dither during playback or when the wordlength are the same. You can hear the final results with dithering while exporting using audible bounce though.


Take care!

Actually, this very issue was tackled in depth a couple of years ago IN THIS THREAD and IIRC there was strong concensus to leave it turned on. WARNING: it's a LONG thread that gets a little HOT in places

ORIGINAL: Jose7822

This is not true. Dithering is ONLY applied during a bounce, freeze or export operation when a bit depth (wordlength) reduction takes place. Sonar DOES NOT dither during playback or when the wordlength are the same.

Hi Jose,

So in Sonar-4, if I have a 16-bit project with a Soft-Sampler that has a mixture of 16 and 24-bit samples, should I have dithering turned on?

If I do have dithering on during playback, is Sonar applying dither to the 24-bit samples and only the 24-bit samples?

Heh heh, might be rather simple questions but as robby and John aluded to, all this dithering has got me dithered.

T.S.

P.S. Just two more things:

1- I believe Triangular dithering is not available in Sonar 4, so I guess you're left with using Rectangular dithering (which is inferior to Triangular but should get the job done).

2- In the thread linked by SteveJL, Ron Kupper talks about a bug that was fixed in Sonar 5.2. This bug involves the ommision of dithering during playback when dithering was enabled, but I'm not sure when this was introduced (hopefully not in version 4).


Take care!

You can create presets.