Can somebody please provide information on what the specific dithering options mean?
The labels “Noise Type 1/2” & “Noise shaping 1/2/3” don’t really give much away.
I think that noise types 1 & 2 are triangular and rectangular PDFs, and shapings 1 to 3 are increasingly biased towards pushing the dither noise into the highest frequencies. Why this has never been stated in the manual or the help, nor even explained in this forum, is beyond my comprehension.
Triangular (1) is better than rectangular for various reasons, and there is no need to think about that choice (in the days when computers were struggling to do this kind of work, rectangular was offered because it is slightly easier to calculate, and so used less cpu power).
Noise shaping is harder to choose. The strongest noise shaping (3) is the least audible, but can actually increase noise if the result is processed again - it should only be used for a final master copy; also those few people with the best extreme HF hearing might find it somewhat intrusive. The weakest noise shaping (1) gives the highest but also the most uniform noise; given that dither noise is actually well below the background noise of most venues and microphones (for distant recording of the type I do), I see no reason to use stronger noise shaping.
Paul
PS: Here is a good article on Dither and Noise Shaping.
Thanks Paul, that’s very helpful in the somewhat bizarre absence of any kind of official documentation.
A follow on question if I may; do you know whether the use of the Crystal Resampler eliminates the need to use dithering, if you’re originally working with 32 bit files?
Absolutely not. Dithering is required any time that the bit depth is reduced (and at no other time). PG is very clear that in WaveLab no process is performed automatically - when dithering is required, you must specify it.
Paul