What is EQ?

EQ in plain English

Every sound is a blend of frequencies. The thump of a kick drum is mostly energy below 100 Hz, while the click of the beater hitting the skin sits up around 2 to 5 kHz. An EQ changes the balance between those parts. The bass and treble knobs on a car stereo are an EQ with two bands; a mixing EQ does the same job with far more control, because you choose which frequencies to change and by how much.

The name comes from telephone and broadcast engineering. Long cables dulled the high frequencies of a signal, and equalisers were built to undo that loss so the sound coming out matched, or equalled, the sound going in. In music production EQ gets used in both directions: to fix problems and to change the tone of something on purpose.

Frequency is measured in hertz (Hz), the number of times a second a wave repeats. Young, healthy ears hear from about 20 Hz to 20 kHz (20,000 Hz), and the top of that range drops as you get older. Each doubling of frequency is an octave, so 110 Hz to 220 Hz is one octave and 5 kHz to 10 kHz is another. Your ears hear those two steps as the same size, which is why every EQ screen, including the one in the lab, gives each octave the same width.

Reading the lab

The big screen is a picture of the sound's frequencies, called a spectrum. Low frequencies are on the left and high ones on the right. The pale line is the spectrum of the original loop, averaged over the whole thing, and the teal shape is what comes out of the EQ. While the loop plays, the teal shape follows the sound live.

On top of that sits the pink EQ curve. The middle line is 0 dB, where the EQ leaves the sound alone; anything above it is a boost and anything below is a cut, with grid lines every 6 dB. Each numbered point is one band. Drag a point sideways to change its frequency and up or down to boost or cut, and the band's own shape is shaded in its colour so you can see what it covers.

Under the screen is a strip that splits the spectrum into six named ranges, and each one has a short description if you hover over it. Click a range and the lab plays only that part of the sound. The Listen to this band button does the same for whichever band is selected, following it as you move it, which makes it the quickest way to learn what 300 Hz or 5 kHz actually sounds like.

The waveform screen below shows the whole loop before and after the EQ, rendered through the same filters you're hearing. The panel beside it describes what your last change did, with numbers measured from the loop. Press A/B to switch the EQ off. The lab sets the original to the same loudness when you do, so you're comparing tone with tone.

You can swap our loops for your own sound. Upload a file, or press Record and sing or play something for eight seconds, and it loops through the lab instead. Nothing leaves your device.

The frequency spectrum, range by range

The edges between these ranges are rough, and different engineers draw them in slightly different places, but what lives in each one stays the same. Once you know that, most EQ decisions become a question of which range a problem is in.

Sub bass: 20 to 60 Hz

The lowest notes in music live here, such as the long tail of an 808 or the bottom octave of a piano, whose lowest A is 27.5 Hz. You feel this range in your chest as much as you hear it. Club systems and good headphones play it; phone and laptop speakers mostly can't, so a bass line that lives only down here can vanish on them.

Below the notes is rumble, from traffic outside or someone knocking the mic stand. It's often too low to hear on small speakers, but it still uses up headroom, which is why high-pass filters get used so much.

Bass: 60 to 250 Hz

This is where bass lines put most of their weight, along with the body of the kick and the low notes of guitars and keys. It's also the chest of a voice: the fundamental of a typical male speaking voice sits around 85 to 180 Hz. Get it right and a mix feels warm and solid; with too much, it turns boomy and the kick and bass start fighting over the same space.

Low mids: 250 to 500 Hz

Nearly every instrument has some energy here, which is why it causes trouble. Stack guitars, keys, pads and a vocal and their low mids pile up into what engineers call mud, a cloudy thickness that makes the parts hard to tell apart. That's why you'll so often see a cut of 2 to 4 dB somewhere around 300 Hz.

Mids: 500 Hz to 2 kHz

A lot of an instrument's character sits in the mids, like the honk of a saxophone or the nasal edge of a voice around 1 kHz. Small speakers reproduce this range well, so the mids decide a lot about how your mix sounds on a phone. Too much makes things sound honky and boxy. Old landline telephones carried only about 300 Hz to 3.4 kHz, and that's why a voice filtered down to the middle of the spectrum sounds like a phone call.

Presence: 2 to 6 kHz

Human hearing is most sensitive in this range, roughly 2 to 5 kHz, so it's where small changes make the biggest difference. The consonants that make words understandable live here, along with the crack of a snare and the click of a kick. A boost makes a sound come forward and feel clear. Push it too far and it turns harsh and tiring to listen to.

Air: 6 to 20 kHz

Cymbals and breath live up here, along with the shimmer on top of a whole mix. The s and t sounds in a vocal, called sibilance, usually sit between 5 and 10 kHz, and above about 10 kHz is what engineers mean by air: a light, open quality you notice mostly when it's missing. Boosting it makes a recording brighter; too much brings up hiss and makes every s sound sharp.

Cutting vs boosting

An EQ only has two moves. A cut turns a range down, which takes away whatever is in the way. A boost turns a range up, which brings out something the sound already has. That last part trips people up: EQ can only change frequencies that are already in the recording, so boosting 10 kHz on a dull, muffled bass part mostly raises the hiss.

A good rule of thumb is to cut narrow and boost wide. A narrow cut can remove one ringing frequency, such as the boxy tone of a small room, and leave everything around it alone. A narrow boost does the opposite of what you want, picking out a single frequency and making it whistle, while a wide, gentle boost of 2 or 3 dB sounds natural.

Cuts also make room. When two sounds share a range, the louder one covers up the quieter one, which is called masking. Cutting 3 dB at 300 Hz in a rhythm guitar can let the vocal through without touching the vocal at all, and that's often a better fix than turning the vocal up.

Boosts add level. A 6 dB boost doubles the height of the wave in that range, which uses up headroom and can push a track into clipping. It also makes the sound louder overall, and louder nearly always sounds better for the first few seconds. Ears hear more bass and treble in loud music than in the same music at a lower volume, so a boost gets a head start in any comparison. The A/B in the lab removes that head start by matching the levels.

The controls

Ableton Live's EQ Eight and Logic's Channel EQ give each band the same few controls, and so does almost every other parametric EQ.

Frequency

The frequency control chooses where a band works. On a bell it's the centre, on a shelf it's the corner where the shelf starts, and on a filter it's the cutoff. Because the spectrum is measured in octaves, moving a band from 200 Hz to 400 Hz is the same size of step as moving it from 2 kHz to 4 kHz. In the lab, drag a point sideways or turn the Freq control.

Gain

Gain sets how far the band turns its range up or down, in decibels. A 3 dB boost makes the wave in that range about 1.4 times as tall, 6 dB doubles it and a 6 dB cut halves it. On most sources a 1 dB change is subtle and 3 dB is easy to hear, so tone shaping usually happens in moves of 1 to 4 dB. Bigger numbers are for fixing problems or for obvious effects.

Q

Q sets how wide a bell is. A high Q is narrow and a low Q is wide. For a bell at 1 kHz, a Q of 1.41 covers about one octave, from 710 Hz to 1.41 kHz, measured between the points where the boost has reached half its height. A Q of 4.3 shrinks that to a third of an octave, from about 890 Hz to 1.12 kHz, and a Q of 0.67 widens it to two octaves, from 500 Hz to 2 kHz.

Most musical boosts and cuts use a Q between about 0.7 and 2. Narrow settings of 5 or more are for surgery: finding one resonance, a ringing drum or a room tone, and taking it out. Some EQs label the same control bandwidth and measure it in octaves, which runs the other way, so a bigger number means a wider band.

Bells, shelves and filters

Each band also has a type, which decides the shape of its curve. The lab has the six you'll find on almost every parametric EQ.

Bell

A bell, also called a peaking band, boosts or cuts a range around its centre frequency and leaves the rest alone, so on the screen it looks like a hill or a valley. It's the most flexible type and the one you'll use most, because it can be a wide tonal boost or a narrow cut aimed at one problem.

Low shelf and high shelf

A shelf turns everything past its corner frequency up or down by the same amount. A high shelf at 8 kHz with 3 dB of gain lifts 16 kHz by nearly the full 3 dB and 12 kHz by a little less. A low shelf does the same at the bottom end, which is what the bass knob on a stereo does. In the lab, as on many EQs, the shelf's frequency is the point where it has reached half its gain, so a 4 dB high shelf at 9 kHz is about 2 dB up at 9 kHz and the full 4 dB above that.

Shelves suit broad tone changes: a little more air on a vocal, a little less low end on a guitar. A bell would put a peak in the middle of the range; the shelf keeps going all the way to the edge.

High-pass and low-pass filters

A high-pass filter lets the highs pass and removes everything below its cutoff, so engineers also call it a low cut. A low-pass filter does the reverse and is also called a high cut. At the cutoff itself the level is already 3 dB down, and past it the level keeps falling at a rate called the slope, measured in decibels per octave.

With a 12 dB per octave high-pass at 100 Hz, 50 Hz is about 12 dB down and 25 Hz about 24 dB down. Switch to 24 dB per octave and those become 24 and 48 dB. A gentle slope sounds natural and leaves a little of the low end behind; a steep one clears it out more completely and can start to sound like a wall. The filter slopes lesson goes further into that choice.

Filters have no gain control, because they only ever cut. Their Q works differently too: on a high-pass or low-pass, turning it up adds a bump just before the cutoff, called resonance. Synth filters use that bump on purpose; mixing filters are usually left at a Q of about 0.7, where there's no bump.

The high-pass and low-pass filters lesson covers where to set them on each instrument.

Notch

A notch is a very narrow, very deep cut that removes one frequency and almost nothing either side of it. The usual job is mains hum: electrical interference at 50 Hz in the UK and Europe, or 60 Hz in North America, often with more hum at whole-number multiples of it. Since a notch is so narrow, you can rarely hear it working on music, which is the point. There's more in the notch filter lesson.

How to EQ, step by step

This order works on almost any sound, from a single vocal to a full drum bus.

1. Listen with the whole mix playingJudge EQ in context. A guitar that sounds thin on its own may sit right once the bass and drums are under it.
2. Name the problem before you touch anythingBoomy, muddy, honky, harsh or dull each point to one of the ranges on the lab's strip, so you already know roughly where to look.
3. Find the exact frequency with a sweepSet a bell to a narrow Q of 5 to 8, boost it by 8 to 10 dB, turn your speakers down a little and sweep it slowly across that range. When the problem jumps out, stop. The lab's Listen to this band button is a gentler way to do the same search.
4. Turn the boost into a cut and widen itFlip the gain to a cut of 2 to 6 dB, then lower the Q until the fix sounds natural.
5. High-pass what doesn't need low endRaise a high-pass filter until the sound starts to thin out, then back it off a little.
6. Leave tonal boosts until lastKeep them wide and small, usually 1 to 3 dB.
7. Compare at the same levelSwitch the EQ off and on with the output gain set so both versions are equally loud. If you can't say what improved, undo it.

How to EQ a vocal, a kick and a bass

Vocals

Start with a high-pass at 80 to 100 Hz to clear out rumble and breath pops. If the voice sounds thick, look for mud between 200 and 400 Hz and cut 2 to 4 dB there. A nasal or honky tone usually sits between 800 Hz and 1.2 kHz. For clarity, a wide boost of 2 to 3 dB around 3 to 5 kHz brings the words forward, and a high shelf of 1 to 3 dB from about 10 kHz adds air.

Watch the s sounds as you add top end. If they start to spit, a static EQ cut at 6 kHz will dull every word to fix a few, so use a de-esser, which cuts only when the sibilance is loud. The de-essing lesson shows how. To hear this in the lab, pick Vocal and solo the presence range, then the air.

Kick drum

The weight of a kick is between about 50 and 80 Hz, and its exact frequency depends on how the drum is tuned, so sweep to find where it sits. The boxy, cardboard tone is around 250 to 500 Hz, and a cut there usually makes a kick sound bigger. The click of the beater is between 2 and 5 kHz, and a small boost there helps it cut through on small speakers. A high-pass at 20 to 30 Hz removes rumble nobody can hear.

Bass

The fundamentals of most bass lines sit between about 40 and 100 Hz; a bass note of A1, for example, is at 55 Hz. That's the weight, but it's also the part small speakers can't play, so the bass also needs something in the 700 Hz to 1.5 kHz range, where its growl and the sound of the strings or the synth's tone let you hear the notes on a phone. A cut around 200 to 300 Hz keeps it from clouding the low mids.

The kick and bass share the bottom of the spectrum, so give each one its own spot. If the kick's weight is at 60 Hz, a small cut in the bass at 60 Hz and a boost at 100 Hz lets both through. The kick and bass lesson covers this and sidechain ducking together.

Starting EQ settings for common sounds

These are places to begin looking. Every recording is different, so sweep around each number and let the A/B decide.

Common EQ mistakes

Boosting until it sounds louder

Nearly every boost makes a sound louder, and louder fools your ears into hearing better. After a few minutes of small boosts on every band you can end up with a curve that's 4 dB up everywhere, which is just a volume change with extra phase shift. Pull the output gain down to match and listen again.

EQing everything in solo

A soloed snare with a big low-mid cut can sound thin and wrong, then sit perfectly in the mix. Use solo to find a problem and the full mix to decide how much to fix it.

Big, narrow boosts

A 10 dB boost at a Q of 8 makes a sound whistle at one frequency, which is what step 5 of the lab plays. Narrow settings are for cuts. If a boost needs to be big to be heard, the sound probably doesn't have much in that range to bring out.

High-passing everything too high

Filtering each track a little is sensible. Filtering every track at 200 Hz takes the warmth out of the whole mix, because each part's low end adds up. Set each filter by ear, and leave it off sounds that carry the low end.

Fixing the wrong sound

If a sample is too bright or too thin for the track, a different sample often beats 12 dB of EQ. EQ is good at moving a sound a little; it's poor at turning it into another sound.

Training your ears to hear frequencies

Knowing that mud lives at 300 Hz is useful, but being able to hear it is what speeds up mixing. The quiz under the lab boosts one of five ranges by 9 dB and asks you which, with the screens covered and the level matched so loudness gives nothing away. A/B stays available, so you can flip between the boosted version and the original as often as you like. Your best score out of eight is saved in your browser.

Between rounds, use the range strip. Clicking through all six on a few different sounds, a vocal, a full mix and the pink noise, teaches you what each range sounds like faster than reading about it.

Types of equaliser

The lab is a parametric EQ, where every band's frequency, gain and Q can be set freely. It's the type in every DAW and the one this lesson has been about. You'll meet a few others.

A graphic EQ has a row of sliders at fixed frequencies, often 31 of them spaced a third of an octave apart, and is common in live sound. Many mixing desks and older hardware have fixed or semi-parametric channel EQs with only a few frequency choices per band. A dynamic EQ has bands that only cut or boost when that range gets loud, a bit like a compressor working on one slice of the spectrum.

Every normal EQ also shifts the timing of the frequencies near the ones it changes, which is called phase shift. It's usually inaudible, though it can matter when two microphones pick up the same drum. A linear-phase EQ avoids it at the cost of extra delay and a faint smear before sharp hits. For most mixing, a normal EQ is the better choice.

Questions people ask

Keep going