Digital audio is a very different beast to its analogue forefather. Cleaner, colder, trickier. The science behind it is enough to make most people’s eyes glaze over, but there are some important rules that are often broken by those who’ve not taken the time to look into it.
To save you the time, here are the fundamentals of digital audio (without the number crunching).
Levels and clipping
The most common digital mistake is inheriting the ‘run it hot’ mentality of an analogue engineer. Under no circumstances should you run a digital signal into the red. Unless you want your audio to sound edgy and horrible, which you might. (There are always exceptions).
Analogue clipping makes a pleasing distortion that if used sparingly can really enhance a sound, digital clipping is an audio error code. If any digital clipping is to occur, it should be left to the mastering engineer to make it happen. Even then, it’s fraught with controversy, and we avoid it as much as possible.
The rule of thumb when recording is to keep it clean, firmly in the green. If recording in 24 bit, hitting somewhere around -18dB on your VU meter will give you plenty of signal to work with while giving you plenty of headroom for peaks.
The sampling rate for CD quality audio is 44.1khz. Therefore lots of people work at that rate. However, DVD quality is 48khz, presenting a good reason to record at the higher rate.
It is generally a good idea to record and mix at higher than CD quality as any processing you do will degrade the original fidelity of the signal to some extent. Some plugins also work better at higher sampling rates (also why many now internally upsample). The higher the rate the better, so work at the highest rate your processor can handle.
But, keep downsampling to the mastering engineer. Using sub-par converters can cause undesirable effects when converting sample rates, so if you’re not sending it out for mastering, you might be better off sticking to 44.1 all the way through. Check out http://src.infinitewave.ca/ to see where yours stack up (pictured).
CD quality audio works at 16bit (as does DVD) so this is definitely the end goal. However, as with sampling rates, using a higher bit depth during tracking, mixing and mastering will help preserve fidelity and often improve processing quality. There has also been talk of the introduction of 24 bit downloads so it’s a good idea to make sure you’ve got the audio to back it up – you can up the bit rate, but all it’s doing is adding 0s on the end.
Dithering is a fairly counter-intuitive process that involves adding a layer of distortion to smooth out any errors made in chucking away 8 bits of information (when downgrading from 24 bit to 16 bit).
Again, this is something that should be left to the mastering engineer, but if you are mastering at home then make sure you always dither last – that includes after changing sample rate. So, if you recorded in 48khz/24bit and you’re exporting to 44.1khz/16bit you will need to first export at 44.1khz/24bit and then separately dither down to 16bit. Putting a dithering plugin on your post-fader master channel and then going straight to 44.1/16 puts your sample rate conversion before your dithering. Your audio will suffer for it.
Also try out as many dithering algorithms as you can, they do sound different. Noise shaping algorithms generally work best, but it is entirely subjective. My personal favourite is Izotope’s Mbit+ algorithm but I will always audition different noise levels and shapes for each project.
Follow these rules and your sound will come out cleaner, sharper and, well, better.