Red Dog Music | Oct 9, 2018 | 0
How does digital audio work?
Digital audio has been with us for quite a while now, long enough in fact that a good number of people probably haven’t heard anything that isn’t digital audio. There are plenty of articles that go into a lot of (bit) depth (sorry, couldn’t resist) about what digital audio is and how it works, but, for those who are just getting into recording, we thought we’d take a look at things here.
Pictures of digital audio
Perhaps the easiest way to think of digital audio is a series of ‘pictures’ of the sound. If someone in front of you raises a hand to wave, that movement is continuous, but if it were filmed, it would be a series of individual still images. Unfortunately, that’s where our film analogy ends. With digital audio, rather than just playing back these snapshots one after the other, there is some clever maths involved…
The quality of our digital file is determined by two parameters: the sample rate and the word length (also referred to as bit depth). These are terms you may have some across before when looking at the specifications of audio interfaces, or when importing CDs to your computer.
The sample rate (measured in Hz, or cycles per second) determines how many ‘snapshots’ are taken each second, and the word length (in bits) specifies how ‘accurate’ each snapshot is. The following image isn’t a true representation of how digital audio works and the bars are simply superimposed over the waveform to explain the difference between sample rate and word length.
As you can see from the image, as sample rate and word length increase, the sequence of snapshots (the red bars) more closely resembles the wave. As the word length increases, the bars can be a greater range of ‘heights’. With 1 bit (at least in this context, there is another form of 1-bit audio that we’ll discuss in a future post), each bar is just represented by, well, one bit, so can only be a 0 or a 1, so only at minimum or maximum height (first wave). Moving to two bit, the bar can be four heights (second wave).
If we then double the sample rate (third wave), we can see that we’re getting closer to what the analogue wave looks like. At the CD standard of 44100 Hz and 16-bits, there will be 44100 of those ‘bars’ every second, and the 16-bits mean that the bars have 32,768 ‘heights’ available to them.
Bits and pieces
Now, there are a couple of final things to cover in this ‘how does digital audio work’ post. The first is something called the Nyquist Theorem. This states that the sample rate has to be twice as high as the highest frequency you want to capture. As human hearing extends up to (for some!) 20 kHz, that explains the 44.1 kHz choice for the CD standard.
The second is that you might naturally think that we must continually use higher sample rates and longer word lengths to improve fidelity of our digital recordings, but that’s not necessarily true… Current 24-bit, 44.1 kHz or 48 kHz recording may well be perfectly appropriate.
24-bit is always nice as it allows for more headroom while recording. As 24-bits gives you more dynamic range, you can still record with a good signal-to-noise ratio, but not get too close to pushing your meters into the red. This allows you to record without using any compression on the way in, giving you more flexibility when it comes to the mix.
As we mentioned previously, we can hear up to 20 kHz or thereabouts, so recording at 44.1 kHz is sufficient to catch that. Now, there is the argument that there are frequencies above that that we can percieve, but which we can’t hear. Now, the Nyquist theorem states that we need to increase our sample rate to capture those, but how often do you record with ultrasonic microphones?
Some excellent white papers from Dan Lavry go into quite a great amount of detail about why we don’t need to keep pushing and pushing the sample rate envelope. Part of the problem is perhaps the diagrams. Much like the one above.
There is the idea that the bars illustrate digital audio and that to match the curve as closely as possible, you need to keep increasing the sample rate and the word length. This is not the case; the mathematics of digital audio takes care of this. If that fails, try to picture this image: play a song through your speakers from a record and from a CD. Do you think that the speaker cones move smoothly when playing the record and in wee, discrete jumps when playing from the CD? Thought not.
All that said though, even at a given sample rate and word length, the quality of the digital converters perhaps can make a difference, but that’s for another day!
Red Dog Music is the UK’s friendliest musical instrument and pro-audio dealer. Between our 5000 square foot Edinburgh shop filled with an incredible range of products, and our London showroom in Clapham specialising in high-end instruments, dj and pro-audio, Red Dog Music has you covered from north to south and from performance to playback.