When Should You Use Dither?

How often and when dither is best used. Don’t let the hype drive you bananas—we’re talking about a spot of icing on the cake here—the cake will taste just fine without it.
We all want to max out that sweet spot, so when you are processing recordings digitally, you should nevertheless apply dither at the right time. Lately, ever more software bundles and hardware devices offer the functions you need for this purpose.
First of all, you should make a habit of passing on unnecessary processing operations.
For example, if you plan on compressing audio data, there’s no need to normalize it first—you’re best off recording it at the highest possible level anyway. If feasible, use an Eq only at a single position within the signal chain.
As a rule, carry out only the processing operations that you really need—every type of processing colors the audio signal to some albeit minimal extent. Of course, if you like wallow in lo-fi dirt, then process to your heart’s content;
anything that adds grunge to the signal will make you all the happier.
Dither the final master signal post-master fader when you are doing the final mixdown. If at some point you have to convert word lengths, be sure to fall back on dithering. After all, dither can mask the rounding errors of complex processing operations—try it out for yourself and see what happens. In any case, do not use dither everywhere at every opportunity.
Keep in mind that every dither operation adds noise in the form of the dither signal. Besides, extreme noise shaping processes can generate undesirable side effects in the course of further signal processing, for example when you’re cutting, also possibly when you’re compressing data.