>Step by step
Understanding the artefacts of EQ
1 In this tutorial, we’ll expose the often misunderstood side effects of using EQ. We’ve all been blown away at some point by what can be achieved with EQ, but there are downsides. It’s surprisingly easy to ruin the fidelity of a mix with poor EQ choices, without even realising EQ’s causing the problem. If you’re gonna use EQ, you should be aware of its dangers... 2 We’ll first look at how EQ affects waveforms. The waveform of a sound is what the sound is. The waveform is the pattern of vibrating air molecules that our ears interpret as sound. There’s no such thing as a process that will change a sound without also affecting its waveform (although it is possible for two different waveforms to sound indistinguishable). 3 This audio file is made up of one second of silence, with a -6dB spike in the middle that lasts for only one sample. This is literally the shortest event allowed by the sample rate. I’m using a 32-bit floating-point WAV file (also in the Tutorial Files), as a 24-bit file isn’t high-res enough for this. If your DAW doesn’t import 32-bit files, you won’t be able to follow along! 4 Interestingly, our spike has a totally flat frequency response. It is possible to EQ the spike, though. You’ll notice from the frequency analyser that any cut or boost we make with an EQ is reflected with a corresponding peak or trough in the spike’s resulting frequency response. What’s happening to the waveform, though? 5 Let’s make a large boost with the EQ, then render it to a 32-bit file so that we can closely examine the waveform, post EQ. If we zoom in, it appears that our instantaneous spike has gained a sinusoidal tail. Let’s select all of the samples immediately following the initial spike and mute that entire area, to effectively restore our original waveform. 6 Now the peak in the spectrum has disappeared. The implication here is that the peak in the frequency spectrum was only possible due to the addition of a residual ringing, after the event. So what happens if, instead of a boost, we make a deep cut with the EQ, and render again? Under close inspection, the waveform is once again skewed.
7 Evidently, even a cut with an EQ results in a residual tail, albeit a shorter one. Again, if we remove the tail, the resulting frequency display becomes flat. Although we’ve made huge boosts and cuts for this experiment, the fact remains that it’s impossible to alter the frequency spectrum of a sound without having some impact on its waveform. 8 What about linear-phase EQ? Let’s try another large boost on our spike using a linear-phase EQ. We’re using FabFilter’s Pro-Q 2 in Linear Phase mode, but any linear-phase EQ will behave similarly. In this case, although there’s slightly less residual ringing following the initial spike, there’s also a pre-ringing effect that happens
ahead of the spike. It seems no EQ is safe. 9 Now let’s look at a more complex waveform: a sawtooth. This is a superclean one, quite possibly the purest you’ll find in the digital domain. Our oscilloscope will track the waveform and display a single period, allowing us to watch what happens to the sawtooth’s physical waveform, while applying EQ with Pro-Q 2. 10 As you’ll see, whether cutting or boosting, in either the standard minimum-phase mode or in linear-phase mode, adjusting frequencies contorts the waveform. This can have a huge impact on how the elements of your mix sum together. In other words, after you’ve established a rough balance, the use of EQ moves the goalposts. 11 Note that smaller boosts and cuts have less-adverse effects, but there’s always some effect. The trouble is, they’re cumulative – they can gradually creep up on you, resulting in mixes that lack transparency and definition. Realistically, some light EQ here and there isn’t going to kill your mix, but it’s best to keep these facts in mind when you want to go too far. 12 One quick tip for EQing drum hits is to use a linear-phase EQ to process the sample, then to render it and chop off the pre-ringing frequencies. You could even loop the sample while you EQ it and use a plugin such as Cableguys’ VolumeShaper to silence the channel during the phase that the pre-ringing occurs. Check my Pro
Tips above for more thoughts on EQ.