Tutorials Audio

EQ Basics: What Every Control Does

Beginner · ~15 min

Overview

An equaliser turns one frequency range up or down while leaving the rest alone. Every EQ, from the two knobs on a mixer to a thirty-band plugin, does the same job with the same handful of controls. Learn the controls once and every EQ you meet afterwards is familiar. This page covers what each control does to a sound, rather than a list of settings to copy, because the useful setting depends on the recording in front of you.

What You Need

Any EQ plugin, including whichever one shipped with your software
A recording with a problem you noticed by ear
Headphones or monitors you have listened to enough to know
A way to bypass the plugin quickly, because the bypass button teaches more than the graph does

Steps

Learn the four filter shapes

Four shapes cover almost everything. A bell, also called a peak, lifts or lowers a band and returns to nothing either side of it. Use it for surgical work and for gentle tonal shaping alike. A shelf slopes up or down and then stays level, so a high shelf at 8 kHz affects everything above 8 kHz rather than a band around it. A high-pass filter removes everything below a point, and a low-pass filter removes everything above one. A notch is a bell with an extreme narrow setting, for killing a single tone such as mains hum.

Read the filter names correctly, because they are backwards

Filters are named for what they let through, not what they remove. A high-pass filter passes the highs and removes the lows. Plenty of software labels the same control a low cut, which describes the effect instead. The two names mean one thing. This trips people up for years, and reading the name as a description of what survives fixes it permanently.

Understand Q, the width control

Q sets how wide a bell is. Low Q values, below 1, give a broad curve affecting a wide range and sounding like a tone control. High Q values, above 1, give a narrow curve affecting a small range and sounding surgical. Match the width to the job. Removing a ringing resonance wants a narrow setting, because the problem lives at one frequency and everything either side of it is fine. Adding warmth or air wants a broad setting, because tone is a property of a range rather than a point.

Sweep to find the problem, then decide what to do about it

The fastest diagnostic in audio takes under a minute. Make a bell, set a narrow Q, add a large lift of 10 dB or more, then sweep it slowly across the spectrum while the audio plays. Whatever is wrong gets much worse as you pass over it, and the harshness or boxiness will leap out. Stop there. Now pull the gain down instead of up, and widen the Q until the fix sounds natural. You found the frequency by exaggerating it and you treat it by reducing it.

Reach for a cut before you reach for a lift

Both directions are legitimate, and cutting is right more often for two reasons. A lift adds level as well as tone, so a lifted band sounds better on first listen almost regardless of the frequency, which makes your ears an unreliable judge. A lift also spends headroom you will want later. Removing the frequency getting in the way leaves everything else standing where it was, and the part you wanted forward comes forward on its own. Where a lift earns its place, keep it broad and small, two or three decibels across a wide curve.

Use a high-pass filter on almost everything

Most recordings hold energy below the range of the instrument or voice in them. Room rumble, handling noise, air conditioning, footsteps through a mic stand and the proximity effect of a close microphone all live down there. None of it carries information and all of it eats headroom. Sweep a high-pass filter upward until you hear the sound start to thin, then back off. On a spoken voice the filter usually lands somewhere between 70 and 100 Hz. Check the result on speakers with real bass response, because small monitors and laptop speakers will not reproduce what you removed.

Match your gain before you judge the change

Louder wins every comparison. If your EQ move raised the output by three decibels, you are comparing loud against quiet rather than good against bad, and the loud one wins whatever it sounds like. Use the output control to match the bypassed level, then toggle the bypass. Half the moves surviving a level-matched comparison were worth keeping and the other half were volume in disguise.

Fix it at the source when the source is the problem

EQ shapes what you recorded. It does not replace what you failed to record. A voice sounding boxy because the microphone sat in the corner of a bare room stays boxy after treatment, because the room resonance and the voice occupy the same frequencies. Moving the microphone takes two minutes and outperforms any plugin. Reach for EQ when the recording is finished and for the microphone when it is not.

Pro Tips

Bypass constantly. Ears adapt to a change within seconds and start hearing it as normal, so a move sounding like an improvement after a minute of listening often sounds like nothing at all against the bypassed original.
Work at a lower monitoring level than feels natural. Quiet listening exposes balance problems and hides the flattering low end of loud listening.
A narrow lift is nearly always wrong. Narrow settings suit removal. If you find yourself making a sharp narrow lift, the sound needs a different recording rather than a different curve.
Solo tells you what a track sounds like alone. The mix tells you whether it works. Make the decision in the mix, because a track sounding thin on its own often sits perfectly.
Mains hum sits at 50 or 60 Hz depending on your country, with harmonics above it. A narrow notch at the fundamental and one or two harmonics deals with it, but re-recording away from the cable causing it deals with it better.
Trust the sound rather than the picture. A spectrum analyser shows you what is there, not whether it belongs. Plenty of correct mixes look wrong on a graph.
Save your own presets once you find settings working on your voice and your room. Starting from a known point beats starting from flat every session.

Knowledge Base

What You'll Learn

The table below is a reference for what lives where. Treat it as a starting point for a sweep rather than a set of values to dial in, because the frequency causing a problem on your recording depends on your voice, your microphone and your room.

What lives in each frequency range

Range	What lives there	Too much sounds	Too little sounds
20 to 60 Hz	Sub-bass, rumble, room noise	Muddy, uncontrolled	No real effect on voice
60 to 250 Hz	Bass, the body of a voice	Boomy, boxy	Thin, weak
250 to 500 Hz	Lower mids, warmth	Muddy, cardboard	Hollow
500 Hz to 2 kHz	Midrange, vowel sounds	Honky, telephone-like	Distant, scooped
2 to 4 kHz	Presence, consonants, intelligibility	Harsh, fatiguing	Dull, hard to follow
4 to 8 kHz	Sibilance, attack, detail	Sharp, spitty	Soft, veiled
8 to 20 kHz	Air, sparkle	Brittle, hissy	Closed in, lifeless

Two ranges do most of the work on a spoken voice. Around 200 to 400 Hz is where boxiness collects, and around 3 kHz is where clarity lives. A small reduction in the first and a small lift in the second improves more voice recordings than any other pair of moves.

The four filters, and what each is for

Filter	Shape	Best at
High-pass (low cut)	Removes everything below a point	Rumble, handling noise, proximity buildup
Low-pass (high cut)	Removes everything above a point	Hiss, taming brightness, effects
Bell (peak)	A band, returning to zero either side	Everything else, wide or narrow
Shelf	Slopes then levels out	Broad tone changes at the top or bottom
Notch	An extremely narrow deep cut	Mains hum, a single ringing tone

A shelf and a wide bell sound similar at first and differ at the extremes. A high shelf at 10 kHz lifts everything above it, including hiss at 16 kHz. A wide bell centred at 10 kHz lifts around 10 kHz and returns to nothing by 16 kHz. When brightness is wanted without the noise coming with it, the bell is the safer choice.

Subtractive and additive, and when each applies

Subtractive EQ means removing what is in the way. Additive means adding what is missing. The first is the default for a reason: nothing in a recording is improved by adding a frequency the microphone never captured, while plenty is improved by removing a frequency masking something else.

The practical version is a sequence. Remove the rumble with a high-pass filter. Find and reduce the one or two ranges causing the problem you noticed. Listen again in the mix. Only then consider whether anything is genuinely missing, and if it is, add it broad and small.

The exception is a recording made deliberately flat. Some microphones and many field recorders capture a neutral signal expecting shaping later. There, a lift is restoring an intended sound rather than inventing one.

Where This Fits

This sits under Mixing a Full Song, the pillar covering the mix process end to end. EQ Cheat Sheet for Vocals gives starting frequencies once the controls make sense, Fixing Muffled Audio applies the sweep technique to one common failure, and Voice EQ for Content Creators covers the spoken-voice case for camera.

FAQ

Q: What does an EQ do?
A: It turns one range of frequencies up or down while leaving the rest alone. Every equaliser does this, from two knobs on a mixer to a thirty-band plugin. The controls differ in number rather than in kind.

Q: Should I cut or lift?
A: Cut first. A lift adds level as well as tone, so it flatters whatever you boosted and misleads your ears, and it spends headroom. Removing the frequency in the way usually achieves what a lift was reaching for. Where a lift belongs, keep it wide and small.

Q: What is Q?
A: The width of a bell. Values below 1 give a broad, gentle curve suited to tonal shaping. Values above 1 give a narrow curve suited to removing a specific resonance. Match the width to whether the problem is a range or a point.

Q: Why is it called a high-pass filter when it removes the low end?
A: Filters are named for what they pass rather than what they remove. A high-pass filter passes the high frequencies through and removes the low ones. Some software calls the same control a low cut, which describes the effect instead. They are the same control.

Q: How do I find the frequency causing a problem?
A: Make a bell, set a narrow Q, add a large lift, and sweep slowly across the spectrum with the audio playing. The problem gets dramatically worse as you pass over it. Then pull the gain down instead of up and widen the curve until it sounds natural.

Q: Where should I put a high-pass filter on speech?
A: Sweep it upward until the voice starts to thin, then come back down a little. Expect somewhere between 70 and 100 Hz for a spoken voice, though it depends on the voice and the microphone. Check on speakers with real low end before committing.