The equaliser is probably the tool you reach for most, and yet it’s one of the most misused. People treat it like a “make it better” knob, when an EQ does exactly one thing: change the level of some frequency areas relative to others. Everything else — fixing a flaw, carving space, adding character — flows from that single operation, provided you understand what each filter type really does to the signal.
What an EQ controls: level, frequency, width
Every band of an equaliser has three parameters. Frequency sets where on the spectrum you act, in hertz. Gain says how far you lift or cut that area, in decibels. And width, usually labelled Q, describes how wide the affected area is around the centre frequency: a high Q targets a narrow, surgical band, a low Q acts broad and musical.
Build this reflex: a tight Q fixes a specific problem (a resonance, a sibilant, a hum), a wide Q colours and adds character. A lot of mixes sound “worked over with tweezers” because everything is done at a tight Q, including the moves that should stay broad.
The main filter families
Beneath the variety of interfaces there are only a handful of filter shapes. Know them and you’ll instantly know which tool to grab.
- The bell (peaking): lifts or cuts an area centred on a frequency, symmetrically on both sides. It’s the do-everything filter of a parametric EQ, with adjustable Q.
- The shelf: raises or lowers everything above (high shelf) or below (low shelf) a frequency, without returning to the original level. It’s the tool for broad, gentle moves: adding air with a high shelf around 10-12 kHz, taming an over-full bottom end with a low shelf.
- High-pass and low-pass filters: they pass one side of the spectrum and roll off the other along a slope in dB per octave. The high-pass filter is the first cleanup move in any mix: pulling useless low end off a voice or a guitar frees up a surprising amount of room.
- The notch: an extremely narrow, deep bell built to kill one precise frequency — mains hum at 50/60 Hz, a room resonance, an emerging feedback tone.
Correcting or shaping: two moves, two mindsets
There are two ways to use an EQ, and confusing them is the source of most mistakes. Subtractive correction removes what’s in the way: a resonance poking out, mud in the low mids, harshness around 3 kHz. You hunt the flaw by sweeping a boosted, tight-Q bell, then attenuate it. It’s a discreet move that cleans the sound without anyone hearing “the EQ”.
Additive shaping adds character: a high shelf for openness, a broad low bump for weight. This is where you colour. The common-sense rule is cut narrow, boost wide — because the ear forgives a targeted cut far more readily than a surgical boost, which always reads as an artefact.
Carving space: EQ in service of the arrangement
The biggest favour an equaliser does isn’t beautifying a solo track, it’s making several sources coexist. Two instruments sitting in the same frequency area mask each other, and pushing the fader only makes it worse. The fix is to gently dip in one what you want to feature in the other: a small mid dip in a keyboard pad to let the voice breathe, a high-pass on anything that doesn’t need low end to clear room for bass and kick. That complementary logic is the same one behind sidechaining, which solves the same clash in time rather than in the spectrum.
Phase, dynamic and linear-phase EQ: the traps
A minimum-phase EQ, analogue or digital, introduces phase shift around the frequencies it works on. That’s not a flaw — it’s part of what makes some equalisers musical and “alive”. Linear-phase EQs preserve phase but generate pre-ring and latency — precious at mastering, often needless and CPU-hungry on every track of a mix.
Dynamic EQ deserves its own place: the band only acts when a frequency crosses a threshold. It’s the ideal tool against a sibilant that only flares now and then, or a resonance that only jumps out in the loud parts, where a fixed bell would harm the rest of the time. It sits on the border between equalisation and compression, and it’s often what you should reach for when a static EQ “over-corrects”.
A method that holds up
In practice, a simple order stops you going in circles. Clean first with a high-pass on anything that shouldn’t reach down low. Then hunt and attenuate the one or two resonances that bother you, on headphones and on speakers. Only colour afterwards, with shelves and a wide Q, A/B-ing constantly against bypass. And keep gain staging in mind: a 6 dB boost changes the overall level and skews the comparison, so trim the output gain to judge at matched loudness.
My take
After years behind consoles where the tone control was often a plain three-band Baxandall, I’ve kept one conviction: the best mixes are almost never the ones with the most EQ on them. Going fully in-the-box gave us parametric EQs of insane precision, thirty linear-phase bands if you like — and that’s exactly the trap. When you can fix everything, you fix everything, and you end up polishing a sound until it’s inert. The question to ask before opening a band isn’t “what can I improve?” but “what actually bothers me?”. Often the answer is one high-pass and one cut. To go further on complementary processing, see reverb, mid/side processing and, on the hardware side, our pick of processors.
