Dithering is a process that helps keep digital audio sounding great, even when some data gets taken away.
Chances are you’ve seen this word when exporting a song from your DAW, but you may be unsure what it means, or if it even makes sense to use it. In a nutshell, dithering helps us maintain the quality of digital media like audio, images, and even video by adding a small amount of noise throughout the entire file. But why would you want to do something like that? Let’s find out, and learn some more about how digital audio works in the process.
Sound as numbers
To really understand the importance of something like dither, we have to first take a step back and talk about how digital audio works. When you record audio into your computer or some other digital device, that continuous wave of analog sound is being sampled and stored as a whole bunch of numbers. This is called analog-to-digital (ADC) conversion, and creates an accurate representation of the original sound.
The amount of space this data will take up depends on sample rate, or how many times per second we want to measure the amplitude level of a sound, and bit depth, or how many discrete values we want to use to measure that amplitude. Bit depth determines how accurately we can represent the level of an incoming sound wave each time it’s being sampled in a process called quantization. Increasing both of these parameters will net you a higher-quality digital recording with more dynamic range, but reducing bit depth later on means getting rid of some information and can lead to something called quantization error.
Dithering: the lesser of two errors
Let’s say you recorded some audio at 32- or 24-bit resolution, but you want to export it to a lower bit rate. Your DAW will perform quantization again, except now it has way fewer values to work with. What ends up happening is a bit like rounding a bunch of numbers up or down, which in this case can lead to quantization error and even audible distortion, once the sound is reproduced through speakers.
This is where dithering works its magic – by adding a bunch of random variations throughout the entire piece of audio in the form of noise, we can effectively reduce the negative effects of quantization error by making it harder for our ears to detect them. You can think about it visually too; a thin pattern of noise across an entire image is okay, because it’s smoothing out the harsher boundaries between lighter and darker values.
How dithering reduces the perceived effects of quantization error
To dither or not to dither?
The ultimate question remains: when is it a good idea to use dither, and when should it be avoided? Here are some tips:
- In general, you should apply dithering any time you’re rendering audio to a lower bit depth (like converting from 32-bit to 24-bit, or 24-bit to 16-bit).
- If you’re exporting at 32-bit file, you usually don’t need to dither, since data isn’t being truncated (removed) during the rendering process.
- If you’re getting a final mixdown or pre-master file ready for professional mastering, you may also want to avoid dithering in case it’s something the engineer would want to perform themselves, but make sure to ask first!
- Don’t apply dithering if you’re converting to .mp3, AAC, or other formats that compress the audio, since this introduces its own artifacts that dither won’t fix.
- You might have the option to select a dithering shape or type. They’re all going to be effective, but feel free to explore each one to see what might work best for your track.
If you have any questions on dithering, let us know in the comments below.
May 18, 2020