Is it better to cut or to boost?

Cut, nearly always, for a practical reason: cutting removes what's in the way without raising the overall level, whereas boosting forces you to pull the fader down afterwards and tends to make you boost the next thing to compensate. That said, a wide gentle boost for air or body is perfectly legitimate. The useful rule isn't «never boost», it's «cut narrow, boost wide».

What is Q and what value should I use?

Q is the width of the bell: the higher it is, the narrower the affected zone. To remove an annoying resonance you go narrow, 4 and up. To add character — more body, more air — you go wide, between 0.5 and 1.5, because a narrow boost sounds like an effect and a wide one sounds like tone.

Should I high-pass every track?

Many of them, yes, but not as routine. It makes sense on tracks that contribute nothing below their lowest note — vocals, guitars, overheads — and there it clears rumble and room noise. Applying it to everything by default, kick and bass included, is one of the fastest ways to make a mix thin.

Do I EQ before or after the compressor?

Depends what you want. Before the compressor, the EQ changes what the compressor detects: take lows out first and the compressor stops reacting to them. After it, the EQ gives final colour without altering the dynamics. The usual arrangement is a surgical cut before and the character boost after.

What's a linear phase EQ and when do I use one?

One that doesn't shift the phase of the frequencies it touches, at the cost of latency and sometimes an audible pre-ring on transients. It's useful in mastering and on parallel material you're going to sum; for mixing individual tracks, an ordinary EQ sounds better more often than people assume.

Why can't I hear my cut?

Usually because you're cutting a little in a zone with a lot of energy, or because the problem isn't on that track. Try exaggerating the cut until it's obvious, confirm you found the right zone, then back off to the smallest amount that solves it. If exaggerating changes nothing, the culprit is another track.

Fundamentals12 min read

EQ: the practical guide that makes sense the first time

What each filter type does, when to cut and when to boost, why subtracting beats adding, and the mistakes that ruin the most mixes.

What an EQ actually does

An EQ turns part of the spectrum up or down. That's all. It doesn't add anything that wasn't recorded and it won't fix a bad take: it redistributes weight among what's already there. Framing it that way avoids half the dead ends, because it stops making sense to hunt in the EQ for a sparkle the microphone never captured.

The second thing worth internalising is that EQ is a decision about the mix, not about the track. A guitar can sound perfect soloed and get in the way the moment the vocal comes in; the cut you make isn't a defect of the guitar, it's room you're handing to something else. Which is why EQing in solo leads to mixes that sound fine track by track and wrong together.

Filter types and what each is for

Filter	What it does	Typical use
High-pass (HPF)	Removes everything below a frequency	Clearing room rumble and footfall on vocals, guitars and overheads
Low-pass (LPF)	Removes everything above a frequency	Cutting hiss and unneeded air; pushing an element further away
Bell (peak)	Raises or lowers a zone centred on a frequency	The workhorse: resonances, body, presence
Low shelf	Raises or lowers everything below a point	Adding general weight without touching the rest
High shelf	Raises or lowers everything above a point	Air and sheen; the most natural boost there is
Notch	Very narrow, very deep cut	Killing mains hum or a single resonance

The method that works: sweep and destroy

There's an old technique for finding the frequency that bothers you and it's still the best one. Take a bell, boost it around 10 dB with a narrow Q and sweep it slowly across the spectrum while the mix plays. At some point the sound turns unpleasant all at once: that's the resonance. Keep the frequency, invert the gain and remove only what's needed.

The mistake nearly everyone makes is leaving the boost in place. With a narrow 10 dB bell everything sounds bad somewhere, so if you go looking for «what sounds odd» you'll always find it, even where there was no problem. The honest check is to bypass the whole EQ and see whether you miss the change; if you don't miss it, it wasn't needed.

Cut narrow (Q of 4 or more): you remove the problem without taking the surrounding tone with it.
Boost wide (Q of 0.5 to 1.5): it reads as colour rather than as an effect.
If a cut goes past 6 dB, suspect the arrangement or the take before the EQ.
Always compare at matched loudness: louder always seems better, and that's how you end up boosting everything.

Cutting to make room, not to fix

Most of an EQ's real work in a mix isn't correcting defects: it's dividing up the spectrum. If the piano and the vocal fight in the mids, nothing is broken in either — two elements are asking for the same place. The fix is deciding which one owns that zone and relieving the other, not EQing both until neither is in the way.

That division goes better when you know what lives where, so it helps to keep the map of the spectrum handy. The particular case of kick and bass, which is where people get stuck most often, has its own guide.

The mistakes that ruin the most mixes

EQing in solo and never re-checking against the full mix.
High-passing every track out of habit, including the ones carrying the weight.
Boosting where something's missing instead of cutting where it's excessive on the track masking it.
Repeating the same move on five tracks: five 3 kHz boosts add up to 15 dB on the bus.
Deciding at high volume, where the ear exaggerates lows and highs and everything seems better than it is.
Not sleeping on it: half an hour later, half of what you did usually turns out to be unnecessary.

To train the part you can't learn by reading — recognising the zone by ear before you touch anything — the EQ Surgeon runs exactly that exercise, and the mix analyser lets you check your judgement against the measurement.

Frequently asked questions

Is it better to cut or to boost?: Cut, nearly always, for a practical reason: cutting removes what's in the way without raising the overall level, whereas boosting forces you to pull the fader down afterwards and tends to make you boost the next thing to compensate. That said, a wide gentle boost for air or body is perfectly legitimate. The useful rule isn't «never boost», it's «cut narrow, boost wide».
What is Q and what value should I use?: Q is the width of the bell: the higher it is, the narrower the affected zone. To remove an annoying resonance you go narrow, 4 and up. To add character — more body, more air — you go wide, between 0.5 and 1.5, because a narrow boost sounds like an effect and a wide one sounds like tone.
Should I high-pass every track?: Many of them, yes, but not as routine. It makes sense on tracks that contribute nothing below their lowest note — vocals, guitars, overheads — and there it clears rumble and room noise. Applying it to everything by default, kick and bass included, is one of the fastest ways to make a mix thin.
Do I EQ before or after the compressor?: Depends what you want. Before the compressor, the EQ changes what the compressor detects: take lows out first and the compressor stops reacting to them. After it, the EQ gives final colour without altering the dynamics. The usual arrangement is a surgical cut before and the character boost after.
What's a linear phase EQ and when do I use one?: One that doesn't shift the phase of the frequencies it touches, at the cost of latency and sometimes an audible pre-ring on transients. It's useful in mastering and on parallel material you're going to sum; for mixing individual tracks, an ordinary EQ sounds better more often than people assume.
Why can't I hear my cut?: Usually because you're cutting a little in a zone with a lot of energy, or because the problem isn't on that track. Try exaggerating the cut until it's obvious, confirm you found the right zone, then back off to the smallest amount that solves it. If exaggerating changes nothing, the culprit is another track.