In order to get through the mindlessly dull, but much needed educational reading of how to dither and why you should do it... I have decided to work as many music puns into this text as I can. I apologize in advance because puns are not my forte.*
"The only thing less exciting than learning about dithering is probably watching a documentary on the history of long division."
I love that quote from the crew over at iZotope. I personally dither with their plugins and will be referencing their tools later in this article. I believe they have a thirty-day free trial if you'd like to follow along. Before we begin, it is best that you know I am only scratching the surface of the science behind dithering. This is a general information article to help your mixes and masters defeat the issues of digital recording. Here we go!
Samples & Bits... the beginning.
When we turn analog sound into digital information (from our favorite microphones/gear, through our interface and into our digital session) we are taking a sound wave with infinite and continuous markers and limiting the amount of points used to define the sound wave. This is how we create a digital signal made of 1s and 0s. If you look at the chart below, the top image shows an analog sine wave and the continuous curve it makes. If you look at the bottom image, you will see a strand of dots that represent how often our digital recording instruments are capturing and replicating that same sound wave.
The process in which a sound wave is measured, captured, and recreated is based on two axis. Horizontal and vertical. The horizontal axis represents the sample rate, or how often a sample is taken per second (examples would be 44.1kHz - 48kHz - 96kHz). The higher the sample rate, the better we capture higher frequencies, or frequencies that have shorter sound waves. The vertical axis represents the bit depth (16bit, 24bit, 32bit, etc). The more bits per sample, or bits per second we measure, the clearer the dynamic range of the sound is. This is why some of the more "indoorsy" engineers obsess over word clocks and will spend thousands of dollars to get the best sample and bit rate they can... yeah, thats you Hank.
The Harsh Truth
Since we now know that recording at 48kHz/24bit resolution and higher is going to capture much clearer audio, and take up much more space on our hard drives, we are going to sit back and enjoy editing and mixing the crystal clear tracks in our session. But wait!!! Most media players only play 44.1kHz/16bit files, and we want more than just the people who visit the studio to hear this perfectly engineered song! Not to mention you would have to open up your session every time someone wanted to hear the song!!! The only way to share your song with the masses is to convert your higher resolution song down to a 44.1kHz/16bit file. So lets try to come to terms with this and begin converting our song.
If you were to open your session and bounce your 96kHz/32bit song with your export settings set at a 44.1kHz/16bit file, then yes! You would be able to play that file on digital audio players. However, your DAW will simply throw out the least important 16 bits and your tracks and effects will not sound nearly as clear as your original mix. We don't want the band wondering why they paid so much money for this low resolution song. To fix this... WE DITHER! If you've ever wondered why your high end sounds muddy, your reverb tails get lost, your snare crack loses punch and edge when you export, you might be needing to go back and make sure you are dithering. It may sound crazy that we are taking our original bit setting and throwing out the least valuable bits to fit our file into the limit of the 16bit max capacity that media players can read, but at least we have a way to minimize the damage so to speak. "Dithering is the process of adding a very low level of noise to your mix to help increase the dynamic range, as well as helping to eliminate the effects or distortion caused by truncation through the process of eliminating the quantization error." - Bob Katz, Click on Bobs name for a link to the more scientific and exhaustive explination of dithering. I often read it when I can't fall asleep... works everytime.
We are now paying for noise?
"Wait! What?! I paid thousands of dollars for the quietest preamps and converters and now I am voluntarily adding noise back in?" said anyone learning about dithering. It is true, we are adding noise to the mix. Luckily enough, this added noise is a pleasant sounding noise and set at a very low level that is not in the perceived hearing range when listening to your song. Based on what program you use, the type of noise, and the shaping of noise, you can choose the type and amount that best suits your song.
Here is the trade off plain and simple... when choosing your dithering type and amount, the more dithering you add, the better your song will convert to 16bit, but it also means that more noise will be added to the song. On the other hand, a small amount of dithering will add less noise to your song, but not deliver as clear as a conversion. This is where it is up to you to decide how much noise is worth the clarity.
I would recommend intentionally listening for noise from max dithering and decide whether it bothers you or not. I personally live in the world where tape hiss and ambient noises are often a good thing, and so a little extra dithering noise is the least of my worries. I have only once ran into the problem of a pop artist noticing that my mix did not have the hiss that the master did and they wanted it "as clean as possible." In this case, I bored them to death with my lecture on dithering in hopes of crushing their spirits with relentless scientific data. In the end, I'm a push over with clients and I asked the mastering engineer to lower the dithering amount. But that is another story about perseverance I will save for another time.
The reason I am posting this information is because I care about you. I care about your creativity and I don't want something as uncreative as dithering to get in your way.. Now keep reading!!! You're almost there!!! It will just take a minuet.*
DC Offset awareness and outreach program.
I'm going to talk about DC Offset really quick... it's bad. It takes up useless bits that are needed in the dithering process. Make sure your DC Filter is engaged on your plug-in. Do a quick google search of DC offset if you're still awake after reading this.
The image below is a great starting point for dithering settings (in iZotope Ozone 6). From here you can adjust things as needed. To dive into all the intricacies of this plug-in, I highly recommend reading the manual. There is so many good features it has to offer.
Please do not!
All science aside, this all should only take a couple minutes to hit the dithering buttons when exporting and call it a day. To end this riveting article on dithering, here are some things that should never be done. They will only lead to treble.*
1. Do not dither until the very end. I have received multiple mixes from clients at 44.1kHz/16bit that they recorded at 48kHz/24bit. Please send your mastering engineer the song file in the setting you recorded your song in. Let them worry about the dithering!!
2. Make sure you are dithering in stereo if your song is mixed in stereo. The noise on the left and right signals should not be identical. The two noises should never align. Most plugins will automatically do this if they are assigned to a stereo track.
3. Dither is always at the end. Your master bus should be in this order, 1-Audio, 2-Effects like EQ, compression, and maximizers, 3-Output level. 4-DC offset filter. 5-Dither 6-Output meter.
4. Only dither once. Once a track has been dithered you can't go back! No editing of the dithered track besides trimming the edges. If you made a mastering mistake, go back to the un-dithered session and re-dither your track after the fix has been made.
5. Enjoy your dithered mix! Make sure you share this newly acquired information with your engineering buddies and save lives.
- Kyle Monroe
Tiny Tape Room