So - you've just finished making a track and are excited to render it down. Most DAWs, like Ableton, have an option for something called dithering. Which option should you choose? What is dithering? When is it used? We'll dive into all of these questions and more in the first of our new blog series.
Let's start here - what is dithering? For now let's say it is very faint noise that is added to the master channel of your track. To understand how and why, we first have to understand a bit about sample rates and bit depth.
Analog audio signals are made up of constant electricity. The waveform is a smooth signal that is occurring in real time. When a pure, constant, analog signal is converted into a digital file on your computer, like a .WAV or .aiff, it is impossible to exactly replicate the true signal. Instead, the computer takes millions of super-tiny snippets called samples. Each sample is a small fraction of a second long. If you render your song at 24 bits and 88.2kHz, there are 88,200 samples per second of audio! Each sample can contain 24 bits of information!
This works sort of like frames in a video. When we watch a movie or television show, we aren't seeing a constant stream of images like we do in real life. Instead, we are seeing thousands of still pictures every second. Our brain just automatically connects them together and we see things as one fluid, continuous video.
Now that we know digital audio files are not constant, but instead made up of super-tiny blocks of time called samples, we can begin to understand dithering.
Imagine you are looking at a staircase from the side. Every step in the staircase is one sample of audio. If you were to roll a ball down stairs, it would be a bumpy ride. Now imagine you went to the top of the stairs with a big, red carpet. Dither noise is like rolling this carpet out down the stairs to make a smooth ramp for the ball to roll down. It helps digital audio signals to be reproduced more accurately.
So, when is dithering necessary? Good question. Dithering should be applied whenever the bit depth of a signal is being reduced. Usually, ONLY at the end of mastering. When you record and render your song and prepare your files for the mastering engineer, you will typically be working at a bit depth of 24 bits (24 bits per sample of audio) or 32 bits. Render your song WITHOUT dithering at whatever bit depth you are working with when preparing your premaster.
Only after mastering processes are complete will the engineer re-render your master to the industry-standard 16 bit 44.1kHz .WAV (CD quality) file. This reduction in bit depth results in the truncated quantization error that we have been comparing to a bumpy set of stairs. The 16 bit audio is even more bumpy than the digital audio we started with, since the 'frames' of our 'video' are not as small or abundant, but more choppy. Dithering is applied to smooth this out, preventing the distortion that would occur otherwise.
So, let's wrap things up concisely - if you are producing music, render at a bit depth of 24 bits or higher without any dithering being applied. Dithering should only be applied when we are reducing the bit depth of a digital audio file since lower bit depths cannot reproduce the signal as accurately and are prone to rounding errors. A song should be dithered only once, almost always as the final step in mastering before creating the 16 bit master audio file that is ready for release. So, turn that confusing option on your export audio menu to 'no dither' and worry about it no more!