“What are the consequences of using data compression after dithering? I use dithering and noise shaping when I bounce to AIFF files from ProTools for CDs. To create MP3 files, I have been converting those same AIFF files to MP3s using N2MP3. Can I use a dithered AIFF file to create MP3s without adding artifacts, or should I make a separate bounce with no dithering or noise shaping for that purpose?”
Any time you are going from a given bit depth to a lower one it is appropriate to use dither, and this is no exception. In the words of Sweetwater Sales Engineer, Nika Aldrich, “To not use dither in these instances is to ‘break’ your audio.” You aren’t adding anything extra to the audio that’s going to interfere with a perceptual coding system such as MPEG. Shaped dither, or noise shaping, is another matter entirely though. A signal that has had this type of manipulation done to it can cause problems when processed through a data compression scheme. Some of the more modern noise shaping algorithms are designed with data compression in mind, and thus work just fine, but unless you know that the noise shaping process you use falls under this category, it’s best not to use it.
