Don't forget here either that the effects of dithering are cumulative so only do it once as the last thing you do at the point of reducing bit depth.

Absolutely... 
Dither once at final stage when (if) reducing bit-depth

I've yet to come across something where 'Triangular' hasn't been sufficient for me but I'm thinking now I might re-visit Pow-R 1.

 
Ditto.  I've always stuck with Triangle since it seems to work just fine and because it is the default.   But now I'm curious...
 
 
Edit: Jim, have you ever had reason to try different reverbs for this test?   Just wondering whether the quality of the reverb itself or it's frequency range would have any effect since it is the "source" being used by the dither process.
 

My initial test were done with Perfect Space, but I couldn't find an impulse long enough to really home in on the tail.

So I switched to Epicverb and used one of it's XL presets which gave me a tail of about 20 seconds - easily long enough to really hear the tail disappear.

Thanks, Jonesey.   That'll save me time from trying to find an impulse response for Perfect Space that I do not have. 
 

If you plan on exporting your project with the purpose of mastering or dithering using another software with dithering, then don't dither. In Scott's book, he recommends changing the Dithering to the Pow-r-3 option if it's your final work and you won't import into any other software with dithering. Here is an article by Craig Anderton about dithering http://ggjam.com/music/re...musicplayer_101802.htm Peace

A couple more questions about dithering ....

1. Do you dither when bouncing tracks within a given project?

I use Triangular dither when bouncing to tracks within a project.  I do a lot of bouncing in order to combine composite tracks, or to simply reduce the number of tracks.  My experience has been that you have to dither or the bounced tracks will have audible distortion.

2. For those who have tested different dithers, aside from the listening test, have you also tried to visually compare the waveforms? 

If you hear a difference, you should also be able to see that difference in the waveform (high zoom), comparing it to the original waveform. This could be a 2nd way to prove which dither is best ... kind of like getting a 2nd opinion.

Thanks

1. No. Dither is not necessary when bouncing to the same bit depth. Lack of dither causes very low-level (usually inaudible) quantization noise (when bouncing to a smaller wordlength), but not distortion.

2. You won't see anything by comparing waveforms. You probably won't hear any difference, either. In order to hear different dither algorithms you have to have an extremely quiet recording and turn it way, way up. Do that and you can hear the differences. In normal music, however, it's highly unlikely you'd be able to see, hear or otherwise detect any differences.