Cut when a sound has something in it you don't want, like boom, mud or one ringing note. Boost when a sound is missing something it already has a little of, like the bite of a snare or the air on a vocal. Most sounds that need EQ get a cut or two first and a small boost after.
Keep boosts wide and gentle, 1 to 3 dB, and save narrow, deep moves for cuts. Whichever you use, judge it at matched loudness, because the louder version nearly always wins a quick comparison.
At a glance
| Property | Additive (boosting) | Subtractive (cutting) |
|---|---|---|
| What it does | Turns a range up | Turns a range down |
| Effect on level | Louder, and the peaks in that range get taller | Quieter, never louder |
| Headroom | Uses it up | Leaves more of it |
| Narrow settings | Ring and whistle, easy to hear as a fault | Hard to hear, good for one problem frequency |
| Usual size | 1 – 3 dB, wide | 2 – 4 dB wide, up to 12 dB narrow |
| Can it fix a problem? | Only by covering it up | Yes, it takes the problem out |
| Can it add something? | Yes, if it's already in the sound | Only by turning the rest down |
| Best for | Presence, air, weight, character | Rumble, boom, mud, harshness, resonances |
What is additive EQ?
Additive EQ means turning a range of frequencies up. You reach for it when a sound is missing something: a vocal that sounds dull next to the guitars, or a kick with no click. A boost brings that range forward so you hear more of it.
A boost can only lift what's already there. EQ changes the level of frequencies in the recording and never creates new ones, so a 6 dB boost at 10 kHz on a bass that has nothing up there mostly turns up hiss. If a sound is missing a range completely, you need saturation, which adds new harmonics, or a different sound.
Additive moves are usually small and wide. A bell with a Q under about 1.5, or a shelf, lifts a whole region gently, which sounds like the same instrument with a little more of one quality. In the lab, band 5 is a high shelf, and step 2 uses it to make the vocal brighter.
What is subtractive EQ?
Subtractive EQ means turning a range down. It's how you remove things: low rumble, a boomy note, the cloudy low mids engineers call mud, a harsh edge or one frequency that rings on after every hit. A cut takes the problem out instead of piling more sound on top to hide it.
Cuts are also how you make room between instruments. When two sounds share a range, the louder one covers the quieter one, which is called masking. Taking 2 or 3 dB out of the part that matters less often lets the important one through without touching it. The subtractive EQ lesson goes through every common problem, where to look for it and how much to cut, so this page sticks to how cuts compare with boosts.
Is a boost just a cut with the volume turned up?
Often it is, exactly. Take the lab's high shelf boost: everything above 2 kHz goes up by 5 dB. Now switch it off, cut everything below 2 kHz by 5 dB with the low shelf, and turn the output up by 5 dB. The low end ends up back where it started and the top end ends up 5 dB higher, which is the balance the boost gave you.
On this page's EQ it's the same down to the decimal point. Measured at ten frequencies from 30 Hz to 16 kHz, the two settings come out within a thousandth of a decibel of each other, and their phase response matches too. That's expected. A standard EQ, the kind most plug-ins use unless you pick a linear-phase mode, is a minimum-phase filter, and in a minimum-phase filter the phase follows from the shape of the curve, so two settings with the same curve get the same phase. Rendered through the lab, both versions leave the vocal 0.9 dB louder than the original.
Look closely at the big screen when you press that step. The teal shape, which is the sound coming out, is identical to the boost. The pink curve has the same shape but sits 5 dB lower, because the screen draws the bands and leaves out the output gain. Then take the output gain away: the tone stays the same and the vocal drops to 4.1 dB below the original.
So for broad tonal changes, additive and subtractive EQ are two ways of writing down the same curve. They differ only in level, and level is something you can set with a fader. Press A/B with the output at 0 dB and the lab matches the loudness for you: the cut version sounds as bright as the boost, because it is.
Bells don't have a neat opposite in the way shelves do. A 3 dB bell boost at 4 kHz is the same tone as cutting everything except a slice around 4 kHz by 3 dB, and no single band can do that. The principle still holds. At matched loudness you're hearing the shape of the curve, and a boost or a cut is just the way you chose to draw it. The linear phase vs minimum phase EQ lesson covers the other kind of EQ, where the phase works differently.
Why do narrow boosts sound worse than narrow cuts?
This is where the two part company. A narrow bell boost picks out one slim slice of the sound and makes it louder than everything around it, so every note that touches that slice rings. A narrow cut of the same size takes a slice away, and your ears barely miss it, because the frequencies either side still carry the sound.
This was measured long before plug-ins. Listening tests by Roland Bücklein, first published in German in 1962 and in English in the Journal of the Audio Engineering Society in 1981, found peaks in a frequency response easier to hear than dips of the same size, and narrow dips the hardest of all to hear.
The lab lets you hear it on the vocal. Band 3 is a bell at 1 kHz with a Q of 8, which is just under a fifth of an octave wide. Boosted by 9 dB it adds a nasal whistle and makes the vocal 2.6 dB louder overall. Flip the same band to a 9 dB cut and it's hard to pick out at all, and the vocal drops by just 1 dB.
Those two level figures aren't mirror images. A 9 dB boost multiplies the power in its slice by about eight, while a 9 dB cut leaves about an eighth of it, and no cut can take away more than was there to begin with. The Right now panel shows that the slice around 1 kHz holds 23% of this vocal's energy. Multiply that by eight and the boost adds far more than the cut can remove, which is why narrow boosts move the overall level and the peaks so much more than narrow cuts do.
That's where the rule of thumb comes from: cut narrow, boost wide. A narrow cut at a Q of 5 to 10 can take 6 to 12 dB out of one resonance with almost no other change. When you want more of something, use a wide bell or a shelf and keep the Q low. The EQ Q and bandwidth lesson shows how wide each Q setting is.
Boosting, headroom and clipping
Every boost makes the peaks in its range taller. The lab's vocal peaks at −0.9 dBFS, just under the top of the scale. The 5 dB high shelf in step 2 only raises its average level by 0.9 dB, but it lifts the loudest peak to +0.4 dBFS, which is over the top. The low shelf cut gives the same tone and leaves the peak at −4.6 dBFS, with room to spare.
Inside a modern DAW that overshoot is usually harmless for the moment. Ableton Live, Logic Pro and most other DAWs process audio in floating point, so a channel can go over 0 dBFS without anything clipping inside the session. It catches up with you later. Anything still over 0 dBFS when you export to a 16-bit or 24-bit file gets cut off, and plug-ins that model analogue gear often distort sooner when you hit them harder.
Across a mix, small boosts add up. Ten tracks with a couple of decibels of boost each leave a mix that's louder and closer to the ceiling, with less headroom for the master. Cutting where you can keeps every channel near the level you set with its fader, which is part of good gain staging.
None of that makes boosting wrong. If a boost sounds better, turn the output or the fader down by the same amount and keep it.
Compare at matched loudness
Turn anything up by a decibel or two and most people hear it as better, usually as fuller and clearer. That makes boosts look good and cuts look bad in a quick comparison, since a boost makes the sound louder and a cut makes it quieter. Ears also hear more bass and treble in loud sounds than in the same sound at a lower volume, so a louder version gets a head start at both ends of the spectrum.
Step 4 in the lab shows how far that can go. The low shelf cut with no output gain has the same tone as the boost, but it's 4.1 dB quieter, so if you switch the EQ off and on without matching levels it sounds like a duller, weaker version. The A/B button in the lab sets the original to the same loudness as the EQ'd sound, which takes the volume out of the comparison. In your DAW, use the EQ's output gain to match the level before you switch it in and out.
When to boost and when to cut
Ask what the sound needs. If something is wrong, there's a frequency causing it, and a cut removes it. If something is missing, a boost brings it out, as long as it's in the recording to begin with.
| Job | Usual move | Typical setting |
|---|---|---|
| Clear low rumble | Cut with a high-pass filter | Just under the lowest note |
| Thin out a boomy guitar | Cut | 2 – 4 dB, 100 – 250 Hz, Q 1 – 3 |
| Clean up mud | Cut | 2 – 4 dB, 200 – 400 Hz, Q 1 – 2 |
| Remove one ringing note | Narrow cut | 6 – 12 dB, Q 5 – 10 |
| Bring a vocal forward | Wide boost | 1 – 3 dB, 3 – 5 kHz, Q under 1.5 |
| Add air | High shelf boost | 1 – 3 dB from 10 – 12 kHz |
| Add weight to a kick or bass | Low shelf or wide bell boost | 2 – 3 dB, 50 – 100 Hz |
| Tame harshness | Cut | 1 – 3 dB, 2 – 5 kHz |
| Make the whole thing brighter | Either; pick by level | A high shelf boost, or a low shelf cut and more fader |
Character is the other reason to boost. Some EQs are bought for how their boosts sound, and the Pultec EQP-1A is the best-known example: its low section has separate boost and cut knobs that both work at the same chosen frequency, 20, 30, 60 or 100 Hz. Turning both up at once boosts the bass and dips just above it, which the Pultec low-end trick lesson explains with a working copy.
Making room: cut one part or boost the other?
Suppose a vocal is getting lost behind an acoustic guitar. The additive fix is a presence boost on the vocal, which makes it louder in the range where the words carry, somewhere around 2 to 5 kHz. The subtractive fix is a 2 to 3 dB cut on the guitar in that same range, which makes space for the vocal without changing it at all.
Try the cut first. The vocal stays as it was recorded and the mix gets no louder, while the guitar rarely misses a few decibels of upper mids when it's playing under a singer. If the vocal still needs help after that, a small boost on top usually does it. A kick and a bass line fighting in the low end work the same way, and the kick and bass lesson covers them.
How to use additive and subtractive EQ together
The last two steps in the lab follow this order on the vocal. A wide 3 dB cut at 300 Hz takes 0.6 dB off the level, then a 3 dB bell at 4 kHz brings the words forward, and together they leave the vocal 0.2 dB quieter than the original, with its loudest peak just 0.3 dB higher.
- Listen with the rest of the mix playingDecide whether the sound has a problem, is missing something, or both. A sound that's fine in context doesn't need EQ.
- Cut the problems firstHigh-pass any rumble, then find mud, boom or ringing with a narrow sweep and cut it. Start at 2 to 4 dB on a wide band, and go deeper only on a narrow resonance.
- Listen again before boostingRemoving a problem often makes the rest of the sound seem clearer, and the boost you were planning may no longer be needed.
- Boost small and wideIf something is still missing, add 1 to 3 dB with a wide bell or a shelf. If you need more than 6 dB, the sound probably isn't right for the part.
- Match the level and compareUse the EQ's output gain to bring the level back to where it was, then switch the EQ off and on. Keep only the moves you can hear helping.
Corrective cuts usually go before a compressor, so it doesn't react to boom you were going to remove anyway, and tonal boosts often go after it. The EQ before or after compression lesson covers when to swap that around.
Common mistakes
Choosing by volume
A boost wins a quick comparison because it's louder, and a cut loses because it's quieter. Match the levels before you decide anything, or you'll end up boosting everything a little and wondering why the mix is so loud.
Treating "cut first" as a law
A broad boost and the matching cut with the level made up give you the same sound, so if the boost is quicker to set, use it and pull the output down to match. Where the advice earns its keep is on narrow moves and on headroom.
Boosting to cover a problem
Boosting the highs on a boomy guitar makes the boom feel smaller next to the treble, but the boom is still there, and now the guitar is louder and brighter too. Cut the boom and the rest of the guitar can stay as it was.
Big, narrow boosts
A 9 dB boost at a Q of 8 makes almost anything whistle, as step 5 in the lab shows. Narrow and deep is a setting for cuts and for the boost-and-sweep search, where you boost only to find a frequency and then cut it.
Cutting until nothing is left
Subtractive EQ goes wrong too. Stack enough cuts and a sound loses its body, and then a boost is the right fix. If you've cut the same sound in four places, switch the EQ off and start again with one or two moves.
Training your ears
Both kinds of EQ depend on knowing which range you're hearing. The quiz under the lab boosts one of five ranges by 9 dB with the screens covered and the level matched, and asks which one. Once you can name a boost by ear, you can find a problem to cut the same way, by sweeping a boost until the fault jumps out. Your best score out of eight is saved in your browser.
Between rounds, go back to steps 5 and 6 and switch between the narrow boost and the narrow cut a few times. Hearing how much more the peak stands out than the dip is the quickest way to learn why the two get used so differently.
Questions people ask
Is it better to cut or boost with EQ?
Cut when something is getting in the way and boost when a sound needs more of something it already has. At matched loudness the two can give exactly the same tone. Cutting gets the edge for narrow moves, because a narrow dip is much harder to hear than a narrow peak, and because cuts don't use up headroom.
Does boosting EQ add noise?
A boost raises everything in its range by the same amount, hiss and hum included. So does a cut with the level made back up, which means that at matched loudness neither one adds more noise than the other. The trouble comes when you boost a range that holds mostly noise, like the top end of a dull recording.
Why do people say cut narrow and boost wide?
A narrow boost makes one frequency ring like a whistle, which sounds like a fault. A narrow cut removes a thin slice that your ears barely miss. Wide boosts sound natural because they lift a whole region gently, so that's the safe way to add something.
Is a boost the same as a cut with the volume turned up?
For shelves, yes. A 5 dB high shelf boost at 2 kHz and a 5 dB low shelf cut at 2 kHz with 5 dB of output gain have the same frequency response and the same phase in this lab, and on any EQ whose low and high shelves are mirror images of each other. Bells have no exact opposite like that, but the idea holds: at matched level, what you hear is the shape of the curve.
Can EQ boosting cause clipping?
Yes. A boost raises the peaks in its range, and if they were already near 0 dBFS they go over. Most DAWs work in floating point, so usually nothing clips inside the session, but anything over 0 dBFS gets cut off when you export to a 16-bit or 24-bit file, and some plug-ins that copy analogue gear distort when you drive them harder.
Should I use additive or subtractive EQ on vocals?
Usually both, in that order. Cut first, often a high-pass and a few decibels out of the low mids or a nasal spot, then add a small, wide boost in the presence range around 3 to 5 kHz or a high shelf for air if the vocal still needs it.