Understanding Dither: The Key to Digital Audio Clarity

Dither, a form of noise, is crucial when reducing bit depth to minimize quantization distortion in digital audio processing.

### Demystifying Dither in Audio Production
* 🔊 Dither is essentially noise added to a signal when reducing bit depth to **minimize quantization distortion**, making it **less audible**.
* 🎚 You should apply dither every time you reduce bit depth, including when going from **32 to 24 bits**.
* 🧐 Quantization distortion occurs because digital audio tries to represent analog sources using a **finite number of values**, leading to **errors**.
* 🎲 Dither, being **random noise**, can help by adding enough variation to **preserve the original signal**.
* 📈 It's crucial to apply **dither at the mastering stage**, right before **bouncing or printing your mix to its final resolution**.
* ❗ When using plugins that automatically apply dither, ensure they have a **dither button** or mention it in the **manual**.
* 🔒 If your DAW internally reduces bit depth when bouncing, freezing, or committing to a track, the last plugin in the chain should be **dither**.
* 🙌 Understanding a few key concepts about dither can help you **avoid unwanted artifacts and distortion** in your audio.

you guys asked me so many times to do a video about this, and I listened. Everything you need to know about Dieter. Stay with me because it's actually a lot less complicated than they make it out to be. Big news. I think this is the first video where I'm allowed to announce it. My artist, Bella Kelly's new single video Heartbreak Motel, is going to be out on all platforms on October 8th. Please go watch the video, stream the song, follow the account, and please help me and her with our biggest project ever. Hello everyone, welcome back to Mixbus TV. Hope you're having a great day. Before we start, please check the info box down below for my mixing courses on Pro Mix Academy, free plugin special discounts, and offers. And of course, if you want to support the channel. But most important, if you really want to learn how to mix and master professionally, click the join button down here, become a Mixbus TV member, access the already begun, always growing library of full mixing and mastering courses, start to finish.

special videos exclusive giveaways and a lot more courses are being added, and will be added in the next weeks. Also, you get mixed consultations with me via skype or email. Let's get to it. For this video, I'm gonna lean on the shoulder of a buddy of mine, Ayan Stewart, who wrote an amazing article on dither. So, credit for the great information you're gonna hear goes to him. I will also link his full article in the description down below. Reason why is because he did an amazing job summarizing the subject and explaining a very nerdy and complex thing like dither in a way that is going to be very easy for everybody to understand without me trying to reward his work. So, first of all, what the heck is dither? Well, dither is basically just noise. It's noise added to a signal when changing bit depth, more specifically when we downgrade bit depth to make quantization distortion less audible. It might sound like a lot already, but believe me, it's not that complicated because the things that you

need to know about dither and when to apply it and why we apply it. Do not require you to go into the deep math. Okay, so we'll just give you the straight facts in this video. So, first of all, when do you need to dither? You need to dither every time you reduce bit depth. And yes, that also means from 32 to 24. And we'll see that in a minute. If you reduce from 32 to 24, the strength and the type of detail almost don't count at all. But when you start reducing to 16 bit, a low to medium detail level with some noise shaping, it's probably the best option. And here, I want to quickly jump to what would be the last point of the video otherwise, but there's quite some confusion out there about this specific point, which is: should you use dither when you go from 32 to 24 bit? And the answer is absolutely yes. And actually, this was the subject which sparked this conversation. There's an article published some time ago on B&H that kind

of keeps popping up. And in this article called A Honest Look at the Mystery of the Ether, there's a section that says this: If you will be converting from 32-bit float to a 24-bit fixed, you don't need to dither because 32-bit float only has 24 bits of audio data with eight bits of headroom, in which the 24 bits float. If you'll be converting from 32-bit float to a 16-bit fixed, you should dither because you truncate eight bits otherwise. So it depends, and that is incorrect. And it's very easy to prove. This is actually why me and hearing connected because we agree on this point, and he sent me a quick test that I got to show it to you. Here's an image of a 32-bit float sine wave faded from -132 dBFS to minus 180 dBFS. You see the end of the fade, all normal there. The next image is the same sine wave truncated at 24 bit using flat ether. You can see you can barely make up the noise and the line is

slightly truncated at the end way below what is audible, and that's the correct way to do it using dither when going from 32 to 24. Now let's see if we don't use dither. You see the signal here cuts off abruptly, and you can clearly see the quantization distortion. How this translates in your music to make it simple: every time there is a quiet passage, the tail of a reverb, or a delay, or a fade out, that distortion will be audible, and that abrupt cutoff will be also audible. That's what happens when you don't use dither. If you're liking this so far, we'll go a little more in depth. So why does this quantization distortion happen, and what is it? This is one of the fundamental problems and differences between analog and digital. Basically, we are trying to measure an infinitely variable source (analog) using a finite number of digital values (those ones and zeros). Because of this, there are bound to be some errors. Sometimes the analog level will be a little above, and sometimes it will be a little below.

closest digital value. It's like trying to measure somebody's height using a tape that only displays feet. Sometimes you will have to round up, and sometimes you will have to round down the nearest foot. In digital audio, this error is called quantization distortion. Using 32-bit floating-point renders like all DAWs now have, this level of distortion is so low that it is insignificant. But when you start reducing bit depth and going to 16 bit, that quantization distortion and artifacts and abrupt cuts can be really nasty. Here's a visual representation: 1k sine wave at minus 96 dBFS represented in 32-bit floating point. You see all the points here, all right. Next is the same sine wave but reduced to 20 bit. Without detail, you can see the quantization errors start to be more apparent. And finally, the same sine wave reduced to 16 bit. Again, without either, you can see it start to be a big mess. Now let's not confuse this with the visual representation of the waveform on your DAW because that's just a visual

representation, and also there will be an argument to be made when the sound gets reproduced. All right, so it's one thing one is in your computer, but then it gets reproduced back to and from an analog system, which is your speakers, your headphones. Okay, that's considered analog. So when that happens, it's not as straightforward, but that's a whole different discussion. So when we look at this sine wave as 16 bit, it's a big mess. So how dither can help solve this problem? Well, like we said before, dither is nothing more than noise, and by nature, noise is random. In the early days of digital audio, they figured out that they could use random noise to our advantage. By mixing this low-level random noise, they could add enough variation that the original signal was preserved. See it almost like using masking to our advantage. The key is that the added noise needs to be completely unrelated to the original signal. When this condition is met and the level of dither is correct, not only does it preserve the signal, it actually removes the

distortion that is tied to the frequency content. Let's look at a couple more visual examples. Here's our 1k fade at 32 bits. Notice the tail end that's lower in level than we would expect to be able to capture with 16-bit resolution. Next image, we reduce that signal to 16 bit without either, and this is what happens: the signal drops to zero once it passes -96 dBFS, which is the theoretical limit of the 16-bit audio, and we can clearly see the distortion here. If we add detail before reducing to 16 bit, this is what happens. There are three things to notice here. As Ian says in the article, the original signal no longer abruptly cuts off at -96 dBFS, but instead smoothly fades into the dither noise. Two, this results in a signal-to-noise ratio increase of about 16 dB, so this is how it helps dynamic range. And three, the previously present distortion tones are gone, not masked or buried below the noise, but actually removed. Keep in mind that D3 you only

apply it at the mastering stage, because you keep processing your audio at the highest possible resolution the entire time until you either print or bounce your mix to CD standards, which is 16-bit 44.1. Nowadays, with streaming platforms, we can print or bounce masters high quality, which is higher sample rate and 24-bit, but again, this is important if you've been working on 32-bit float. When you bounce your print to 24, you need to use dither as your very last stage. And I want to add the three addendum that Ian put in the article, because I think they are important. One is self-dither. Some people say, Oh, you don't need to apply dither if you use this or that plugin. Well, make sure, okay, don't go with rumors, unless the plugin has a dither button that you can activate or deactivate, or is clearly stated in the manual that it automatically applies dither, which is not a good thing, because you should be able to decide shape and level. Make sure that you dither every time you reduce bit.

depth. Now, one thing that I should double-check, but I'm 90-90 sure, is if you are on Pro Tools, and every DAW works differently, so check your DAW. When you directly export a file from the project and you decrease bit depth, let's say you have a file, you export it directly from the Edit page. Pro Tools does apply data. Actually, it states that if you bounce and you already applied detail, you will recognize it and avoid to apply twice. Now, I don't know if I trust this. That's not my method. When I print or bounce stamps, I always use a dedicated detail. So, either or, that's not a problem that I personally worry about. The second thing, if you're bouncing, freezing, or printing a complex chain of effects that you have in your DAW. Now, every DAW again works differently, so you should check your manual and make sure that either when you do a processing like that, whether your DAW works internally at 64-32. What happens when you freeze or bounce or commit to a track, okay, and if

your DAW reduces bit depth from 64 to 32 to 24 even automatically. When you do something like that, your last plugin in the chain should be d3. And this would be pretty much it. As you see, it wasn't as complicated as sometimes people make it up to be. Because in the end, there are a few key things that you need to know about Dither for you to avoid bad artifacts and distortion creeping up here and there in your audio. I hope the video was useful. And if you liked it, don't forget to leave a like. Thank again to Ian Stewart. You can find his full article in the info box down below. Why you're there, check all the links. And remember, my artist Bella Kelly's new single Heartbreak Motel is going to be out on our platform on October 8th. Please go watch the video, stream the song, follow the account, and please help me and her with our biggest project ever. Thank you for watching. Subscribe if you haven't already. See you next time. [Music] [Music]

DITHERING DEMYSTIFIED: THE ULTIMATE GUIDE TO DITHERING

David Gnozzi Platinum Mix Mastering Engineer, Producer
My Artist Bella Kelly www.bellakellyofficial.com 1single 750K  https://youtu.be/YT3gXsQ2MsU
2nd at 1M+ https://youtu.be/YuRVDyGL0VY
Book David: bookinghfs@gmail.com
Online mix & mastering services
CREDITS INCLUDE
McSolaar (2x platinum), Ladytron, Felix Da Housecat, Ministry, Raygun Romance, Ryan Shuck (Orgy, Adema, Julien-K), Daniel Graves (Aesthetic Perfection), Wes Geer, Hu3m3n, Brooke Colucci, Mad Gallica, Ty Oliver, System Syn, Danny Blu, Sammi Doll (IAMX), Street Sects, Lizzy Jeff, Vizin, Nolan Neal Nova Melodious Kaos India Hellfire Society Izzy T Robben Ford, Lambstone Rhumornero Colors Void Yng Hstlr Sydney Valette Stonebreed Levysill The Nest Valeria Rossi, Universal, Nettwerk, Danse Macabre, Virgin Radio, Fallout 4, Rock Band 3
David's Premium Courses: https://bit.ly/2RHF5AA
www.mixbustv.com
MixbusTV is a Youtube channel dedicated to professional recording mix and mastering tutorials, music production tips & tricks




MixbusTv

Announcement of Bella Kelly's New Single and Dithering Explained

Bella Kelly's new single "Heartbreak Motel" will be released on October 8th on all platforms, marking a significant project for the artist and the speaker.

Dithering is the addition of noise to a signal when reducing bit depth, and it is necessary every time bit depth is reduced, including from 32 to 24 bits.

The explanation of dithering is simplified, emphasizing the practical application without delving into deep math.

The Importance of Dither in Audio Conversion from 32-bit Float

Converting from 32-bit float to 24-bit fixed doesn't require dither, but it is essential when converting to 16-bit fixed to avoid truncation distortion.

Without dither, the audio signal can experience quantization distortion, which becomes audible during quiet passages or fades.

The use of dither is crucial in digital audio conversion to mitigate the errors caused by the finite number of digital values.

The Impact of Quantization Distortion in Digital Audio

Quantization distortion becomes more pronounced when reducing bit depth in digital audio.

The use of dither, which is random noise, can help preserve the original signal and mitigate the impact of quantization distortion.

Dithering is a technique that adds enough variation to the audio signal, effectively removing artifacts caused by quantization.

Dithering is crucial when reducing bit depth to prevent distortion and improve the signal-to-noise ratio.

It is recommended to apply dither at the mastering stage and when bouncing from a higher resolution to 24-bit.

Self-dithering is important, and it's necessary to use dither whenever reducing bit depth, despite any rumors about dithering plugins.

Best Practices for Managing Bit Depth in Your DAW

When exporting a file from your DAW, be cautious about decreasing the bit depth, as it can affect the audio quality.

It's important to understand how your DAW handles bit depth when bouncing or freezing tracks with a complex chain of effects.

Make sure to check your DAW's manual to see if it automatically reduces bit depth and consider using dither as the last plugin in the chain to avoid artifacts and distortion in your audio.