Buy 'Introducing Dither' video training: £10
When you work on a mix, I expect the furthest thing from your mind will be considering adding noise to your finished project. That's understandable. Why on earth would you spend many hours recording your individual tracks and then possibly spend as much time again mixing your music only to add in some extra noise? Well, that's what this course on dithering will attempt to answer. It will explain the detail behind dithering and why it can be important to add noise, albeit at a low level to your finished mix, prior to converting it to a format applicable to CD playback. Let's start by first defining what 'dither' means. Then, we'll look at how it works to make a mix sound better than a non-dithered mix.
What is 'Dither' and why use it?
Dither is the process of adding very low-level noise to an audio file prior to reducing its sample word size - i.e. the sample resolution. Why include dither? It's done to preserve digital information that would have been lost, or to be more accurate, discarded during the sample rate conversion process from 24 bit audio files, for example, down to CD quality at 16 bit.
What is the consequence of discarding this information from the original mix? The answer is distortion. Whenever you move from one sample rate to another, the process means you add distortion to the resulting 16-bit file. The process of using dither helps to minimise or disguise this distortion. Dithering achieves this by retaining discarded down sampled information. Or, to be more accurate, adding in dither noise assists us by helping to retain more information.
You'll more than likely be mixing your project using 24-bit or even 32-bit audio files. Owing to this, to transfer a finished mix to 16-bit - the only bit rate applicable to CD playback - the conversion process reduces or ‘truncates’ the amount of bits from the original file, in our example, from 24 bit down to 16-bit. It's that truncating process that adds distortion.
Another term for truncating is ‘truncation distortion’ or even ‘quantisation error’. These terms are essentially the same thing. At the very least, this truncation distortion will sound unpleasant. To reiterate, we add dither because it's a process that subtly adds low-level noise that randomises the quantisation errors mentioned. Doing this stops the quantisation distortion from happening when we subsequently alter the bit depth from 24-bit or 32-bit down to 16-bit for a CD master. This is a reason why, if we don't add in the low level dither noise to a mix, the resulting 16-bit file will sound worse.
Let’s investigate what we mean by digital audio. One way to do this is to remember that analogue audio is a continuous connected waveform. By contrast, digital audio might look like a continuous connected waveform, especially when zoomed out. In reality, it isn't continuous. If you look at an analogue audio waveform, we see a continuous smooth flow throughout the waveform shape. However, if we look at a similar waveform in a Digital Audio Editor, and zoom in close to sample level, we see that, although tiny, there is a distance between every single sample point. And each point, in a sense, has a connecting line drawn in to join together those points.
Digital audio isn't actually a continuous sound in the way analogue sound is. Instead, digital audio is recreated by individual bits. We know this as ‘quantisation’ and it attempts, very effectively, to reproduce continuous sound. However, because it's a digital signal, it's simply a vast collection of digital ones and zeros. Zoomed in close to a digital signal shows individual points. From this, we get to see the distance - and by association emptiness - between those points.
We measure digital audio in two connected ways - the sample rate and the bit depth, often known as the word length. We know the sample rate for CD quality playback is 44.1 kHz - or 44,100 samples per second of the audio's duration. We also know the bit depth for the same CD quality is 16-bit. To clarify, the sample rate of an audio file is the amount, or how often, samples are taken. The bit depth used for 16-bit CD quality playback is entwined with the sample rate. So, how do these two qualities measure audio? If we look at a highly zoomed in part of an audio file’s connecting points, the horizontal distance between two adjacent samples measures or represents the sample rate. How often samples get taken and the vertical distance between adjacent samples gives us the bit depth - or word length as it is also known.
At this point, you'd be forgiven for asking ‘Okay! So what? How does that affect or possibly hinder what I hear? That’s what we’ll consider next.
Taken from 'Introducing Dither' training tutorial. (CLICK FOR PREVIEW)
Click the ADD TO CART button above to purchase around 40 minutes of training videos to 'Introduce Dither'.