Dithering

HPF dither
Dither noise is in the high frequencies
"Near Nyquist"
Dither noise is shifted up to near the Nyquist frequency (in a 44.1 kHz recording this would be around 22 kHz). This pushes the noise to the limits of our hearing.
"Psychoacoustic" noise shaping
Using the Fletcher Munson curves and the knowledge that dither noise is very quiet, we can see that the ear is most sensitive (at low volumes) at around 2.5-5 kHz and around 12 kHz. It therefore makes sense to attenuate the level of dither noise at those frequencies, which makes them much less noticeable to us.

How to Use Dithering

  1. Don't apply dither noise to a wave with DC offset
  2. Don't normalise to 0 dB before dithering; leave at least 1 dB safety margin in case the dithering noise pushes the signal into clipping
  3. Don't do anything (apart from final wave edits) after dithering
  4. Only dither an audio file once. Go back to the original 24-bit file if you need another bit depth

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 2.5 License.