What is Dithering Audio?
One of the most obscure questions in the home music studio is related to dithering audio. Look up the word dither on Wikipedia and you'll get a seemingly confusing article. One that may make you feel as though you need a master's degree to understand it.
Let me attempt (and hopefully achieve) to make this simple yet critical process easy to understand. I know some of you enjoy the science behind it all. However, I am going to take a slightly simpler approach as we deal with dithering audio.
When we hear sound naturally, bones in our ears convert the audio vibration to recognizable tones. In digital audio this process is replaced by ones and zeros. The computer is essentially reproducing an audio wave by measuring voltage peaks and valleys. This Digital audio is measured primarily in 2 ways, bits and sample rate.
Bits are the simplest form of this digital information and are needed for any computer to make sense of it all. Bits allow the computer to measure the voltage variances. The more bits available, the more voltage dynamic range there will be. Thus higher bits rates equate to higher quality audio files. There is simply more data within a higher bit rate file.
Sample rate refers to the speed at which the digital audio is processed per second. Again, the higher the sample rate the higher quality the audio file will be. An audio CD has 16 bits of data and is processed 44,100 times per second.
Dithering audio is the process of smoothly converting a higher bit rate and faster sampled file to a lower one.
Imagine you had an audio file recorded at 24 bit with a sample rate of 96000 times per second. Now you need to burn that same file onto an audio CD which only supports 16 bits and sampled at 44100 times per second.
There is information contained in the higher quality file that will simply not be contained in the lower one. Dithering audio attempts to fill in the blanks with step down data or white noise. This makes for a much smoother sounding audio file than simply removing the bits and slowing down the samples from the lower quality file. Without this noise added distortion would occur, something very unpleasant indeed.
What quality should I record at?
There are several factors that must be addressed to properly answer this question. First, your audio device has support for certain bit and sample rates. You cannot record at a greater sample rate or speed than your interface will allow.
Second, the higher the bit and sample rate the larger your audio files will be. Larger files require more computer power to process and more hard drive space to store. This can also affect the latency (speed at which they communicate) between your computer and audio interface.
I recommend trying to capture your tracks at a 24 bits and 48k sample rate. If you have a more robust setup then try using a 96k sample rate. The higher quality you can record and mix at the better your end project will be.
Dithering Audio is best done last
It is always best to keep your tracks at their highest quality as long as possible. Dithering audio is typically the last stage of the mastering process.
Bump your final mix down to the same quality as your individual tracks were recording at. That stereo file should then be mastered without any dithering being applied. Once you have a final mastered audio file it can then be dithered down for various sources (CD, mp3, ect).
Do you have any more thoughts or experience with dither audio? I'd love to hear from you. Please add your input to the section below.
UPDATE:
After a great email discussion with Ian Shepherd from http://productionadvice.co.uk I want to clarify this last statement. Though the title is "Dithering Audio is best done last" I also clarify that by saying "dithering audio is typically the last stage of the mastering process". That word "typically" is what I will explain a bit more. Ian has a great video post that explains the reasons why here. Check that out for more information. The bottom line though is that there are other situations where dithering may be an advantage. Anytime you consolidate your audio tracks into one, or process your audio effects, or export, you are literally creating a new file. I am currently recording with Cakewalk Sonar X1 Producer. Sonar has a default dithering setting which automatically dithers your audio in these situations. During a final mix down, Sonar then gives you the ability to choose your dithering options. It's possible that your DAW of choice does not dither automatically in the situations I've mentioned above, in these cases, you would also want to using dithering.