William Boyle AKA Elfy wrote on Tue, 18 October 2005 10:04 |
High babes. So i sent evaluation files to a guy at 24 bit which i dithered to, from 32 float. I didn't hear from him for a day or 2 until he replied back happy as larry and had dithered down to 16 and sent off for duplication. Since the audio was dithered twice in this manner is it cumulative, or does the first lot of dither get thrown away in the next truncation. Can someone clear this up please. Thanks
|
Dither noise adds, like any other noise. Treat it as noise. The new dither you add is used to eliminate distortion when you reduce the wordlength and it is noise.
The audibility adds up as it accumulates, but you rarely hear it directly, you just hear the effects of it. It's pretty darn hard to hear the effects of even SEVERAL generations of 24 bit dither, so don't think too much about it. It's only when you have more than one generation of 16 bit dither that it becomes iffy.
And then again, when you mix multitrack, the noises of all the individual tracks add far more than any dither issue would. So there is a lot of art and mystery beyond the science here. I'm beginning to theorize that some of the veiling I hear from dithering processes may be due to the calculations of the dither itself, and certainly some dithers sound better (more transparent) than others even when their noise floors are the same.
Glenn Zelniker thinks that the reason POW-R sounds as good as it does is because it computes the random noise and adds it in using floating point calculations. Could be...
BK