I Beg To Dither

When you export the project to audio from a higher bit rate to, say, 16 bits for CD, you need to use dithering to "fill in the bit-gaps". I use Powr-3 IIRC.

I see. It doesn't affect what you do when you are working, just when you export. I've been using power 2, is there any benefit to going to power 3?

You can see the shapes of the dithers if you export silence to a Wave file using the dither, then re-import the wav and normalize it. Then, play it and turn it up so you can hear the noise. You can also but a Freq Analyst on it so you can see where the noise is more and where it is less for the particular algorithm.
 
This one is Power 3.
 

White Noise

I suspected as much. So maybe I'll just choose "none." Why use CPU needlessly? thanks for that, Cactus.

I don't think it's like Dolby. Dolby boosted the very high end of the sound above the noise level on the tape and then when you play it back with Dolby On you suppress the same freq range aligning the original signal and suppressing the high end tape noise.
 
The problem with this was that if you played it back with Dolby off (or the wrong dolby) you could just get extra tape hiss.
 
THIS IS NOT THE CASE WITH DITHER.
 
When you Dither appropriately, you get better sound on all playback environments no matter the settings on the device doing the playback. Dither suppresses the noise floor by randomizing it and spreading it over a greater frequency band (like the audible range). DITHER WILL MAKE YOUR RECORDINGS BETTER.
 
Use Dither when changing Bit Depth (like 32 to 24 or 24 to 16). Don't use Dither when going from 88.2 Samples per Sec to 44.1. I have sources that say going from 48 to 44.1 use dither and others that say don't.
 
http://www.amazon.com/Digital-Audio-Explained-For-Engineer/dp/141960001X/ref=sr_1_2?ie=UTF8&qid=1383132288&sr=8-2&keywords=sweetwater+digital+recording
 
This book says it is appropriate to use dither when changing sample rates where the sample rate is NOT an even division... like 48 to 44.1 but not 88.2 to 44.1.
 
Basically, since it adds the slightest bit of white noise, you want to avoid it where it is not necessary. So, for this reason, I add Dither when going from 32 bit mixdown's to export to 24 bit waves. I usually change sample rate and bit depth at the same time.

OK, I guess I will use it. Actually, I already have been. I just couldn't decide what setting to use, so I tried to use one in the middle, like power one or two. I guess I'll continue doing that. I only got concerned about CPU because the sonar help files mention it.