EQ Settings for Rap Vocals (Cuts First, Boosts Later)

A rap vocal wants two EQs, not one. The first only cuts: high-pass at 80 to 100 Hz on most male voices, 1 to 3 dB out of 150 to 500 Hz where the take is boxy, and one narrow cut of 2 to 4 dB on whatever rings. The second, after compression, saturation and tuning, carries the boosts: 1 to 2 dB wide at 3 to 6 kHz for presence and a 1 to 2 dB shelf at 10 to 12 kHz for air. Cuts before the compressor, boosts after it, and rarely more than five bands in total.

Disclosure: Ghostnote makes audio plugins, and none of them is a channel EQ you can buy, so nothing below is an EQ ad. Two of ours appear, each labeled as ours, because EQ happens inside them, and any measured number credited to us comes from that plugin's probe suite. No affiliate links, and every move here works in the EQ your DAW already ships with.

What EQ settings should you start with on a rap vocal?

Here is the whole cheat sheet in one table. Read it as a map of where to look, not a preset: the only row that applies to every rap vocal is the high-pass, and the only way to know whether the rest apply to yours is to listen with the beat playing.

Q is bandwidth wearing a disguise, and most people set it by feel without knowing what they picked. The conversion is arithmetic: Rane's note on bandwidth in octaves versus Q puts one octave of bandwidth at Q 1.414. Run the same formula outward and Q 0.7 is roughly 1.9 octaves wide, Q 4 is about a third of an octave, and Q 8 is under a fifth. That is the whole reason tonal boosts use Q 0.7 to 1.0 and narrow corrective cuts use Q 4 to 8: one is shaping a voice, the other is removing a single resonance.

Where should you high-pass a rap vocal, and how steep?

Start the corner at 80 Hz for a deep voice and 100 to 120 Hz for a brighter one, at 12 dB per octave, then move it while the beat plays. Below the corner there is nothing you want: stand rumble, footsteps through a wooden floor, the air handler, and the low-frequency lift a directional microphone gives you when the artist is two inches off it. That lift is the proximity effect, and DPA's explanation of it is the clearest short one: it exists on gradient patterns, it is strongest on-axis, and turning the microphone reduces it.

The part almost every cheat sheet skips is that the corner frequency is not the edge of the filter. Take the standard Butterworth response and set a 12 dB per octave high-pass to 100 Hz. It is 3 dB down at 100 Hz, still 5.4 dB down at 80 Hz, and 9.4 dB down at 60 Hz. The same corner at 24 dB per octave is 8.4 dB down at 80 Hz but only 0.2 dB down at 150 Hz. Those are properties of the filter shape rather than a measurement of any one plugin, and your EQ may use a different alignment, but the lesson holds in all of them:

• A shallow filter reaches up. At 12 dB per octave, a corner at 120 Hz is still 1.5 dB down at 150 Hz. On a baritone that is chest, and chest is what makes a rap vocal sound like a person in the room.
• A steep filter is the surgical one. If you need the space under 100 Hz for the 808 and you want the voice untouched at 150 Hz, 24 dB per octave gets you both. Steeper filters also ring more on transients, which is why 12 is the default and 24 is the deliberate choice.
• Set it with the beat playing. Soloed, every high-pass sounds like it is taking too much. In the track, the same filter is usually not enough.

High-passing at 200 Hz because a chart said so is a common way to make a rap vocal thin, and it is worse on deep voices than on bright ones, because the filter is then sitting on top of the fundamental rather than under it. Vowel identity is carried by formants, and for adult male speakers the first of those sits low, in the region people reach for when they want to clear mud. Peterson and Barney's 1952 Bell Labs vowel study of 76 speakers is the classic measurement of where they sit.

One thing a high-pass does not do is fix a plosive. It reduces the thump, but the envelope bump survives and the waveform still lurches on every hard "b". Fix that with clip gain on the word, a steeper filter automated across that syllable, or the artist off-axis at the source.

Why does subtractive EQ go before the compressor?

Because a compressor's detector reads level, and low-mid buildup is level. Leave 4 dB of proximity boom at 200 Hz in the signal and the compressor spends its gain reduction on the boom, ducking whichever syllable happened to carry it. The words that get pulled down are not the loud ones, they are the low ones. Cut the boom first and the same compressor responds to the performance.

There is a second reason, and it is about your ears rather than the plugin. Low-mid buildup makes a take sound darker than it is, so you set the de-esser and the presence boost against a picture that changes the moment you clean it up. That is the order the whole chain follows: cleanup, then control, then color, then shape. The full stage list is in the rap vocal chain in order, where this is stage one.

The mirror image of this is the compressor's own sidechain high-pass, which filters what the detector hears without filtering what comes out. Keep both. Our TALISMAN has one in its ADVANCED panel, feeding both detectors, 80 Hz by default and adjustable from 20 to 500 Hz. In our probe suite, a 60 Hz tone drew 26.3 dB of gain reduction with that filter at 20 Hz and 11.2 dB with it at 500 Hz. On an 808-heavy record, 100 to 120 Hz is the better setting.

The other half of the rule is the one people break without noticing: a boost placed before a compressor is heard by its detector too. Put +2 dB at 4 kHz in front of a compressor doing 4 dB of gain reduction and every syllable carrying that band now pushes harder into it, so the brightest words get pulled down more than the rest. Boosts belong after the dynamics, which is why the chain has a second EQ at all. Settings for the compressor itself are in vocal compression settings for rap.

How do you find the boxy 200 to 500 Hz?

Loop four bars of the verse with the beat playing, then:

1. Insert the EQ first in the chain. One bell, +8 to +10 dB, Q around 5.
2. Sweep slowly from 150 Hz to 700 Hz. You are listening for the spot where the voice turns into a cupped hand or a cardboard tube.
3. Note the frequency. Then throw the sweep settings away.
4. Cut there at 1 to 3 dB with Q between 1.0 and 1.8, and widen until the voice sounds like itself minus the problem.

Step 3 is the one people skip. A +10 dB narrow boost is a detection tool, and it exaggerates on purpose. Inverting it into a −10 dB narrow cut removes far more than the problem, and the result is a voice with a hole in it that nobody can point at.

Two different things live in that range and they want different treatment. Roughly 150 to 300 Hz is chest and proximity, the sound of a close microphone, and it responds to a wide gentle cut. Roughly 300 to 500 Hz is boxiness, which usually comes from the room rather than the voice: a small untreated space, a wall a few feet behind the artist, a desk reflecting back into the capsule. If the ring moves when the artist moves, it is the room, and EQ is treating a symptom. A blanket behind the microphone will do more than any plugin.

Is 400 to 600 Hz mud, and should you cut it?

Mud is a relationship, not a frequency. A vocal that is muddy under a beat with warm keys and a distorted 808 is a vocal that would sound full and correct over a sparse one. That is why deciding this in solo is a mistake: soloed, the cut sounds like damage; in the track, it sounds like the vocal stepped forward.

So run the test both ways. Loop the hook, cut 2 dB at 500 Hz with Q 1.0 on the vocal, and listen. Then undo it and put the same cut on the beat bus instead, with the vocal playing. On most trap and drill records the second version wins, because the vocal was never the thing with too much 500 Hz. Keep whichever clears more, and only one of them.

Numbers that work: 1 to 2 dB on the vocal, Q 1.0 to 1.5, centered anywhere from 400 to 600 Hz. Up to 3 dB on a beat bus carrying keys, pads or a guitar loop. On stacked doubles, cut once on the group rather than once per layer: the same cut on six layers does exactly what one cut on their bus does, and one instance is easier to judge and to undo.

The failure mode to watch for is compound cutting. A 2 dB dip on the vocal, another on the beat bus and another on the master takes 4 dB out of the middle of every part of the record, which sounds clean on headphones and disappears on a laptop.

Why does the vocal sound nasal at 1 to 2 kHz?

Honk sits lower than people expect, mostly 800 Hz to 1.5 kHz, and it is narrow. Sweep it the same way you swept for box, then cut 2 to 4 dB with Q 4 to 8. If the honk is on every single word, it is the microphone or the room. If it appears on one vowel and vanishes on the rest, it is a resonance in that particular voice, and a static cut is the wrong tool: see the dynamic EQ section below.

Now the part that costs people more than the honk did. The 1 to 2 kHz band also carries the forward edge of a vocal, the part that puts a voice in front of a beat rather than behind it. A wide 3 dB cut there, applied because a chart labeled the region honky, is the fastest way to push a rap vocal into the background. Then comes the compensation: a presence boost at 5 kHz to get it back, which buys harshness instead of forwardness, and a de-esser to fix the sibilance that boost created. Three moves to undo one wrong one.

The tell for a cut that went too far here is hollowness. A nasal vocal sounds like the singer is pinching their nose; a hollow one sounds like a phone call. If you are hearing the second, you overcorrected.

Where is presence, and how much can you boost?

Between 3 and 6 kHz, and less than you think: +1 to +2 dB on a bell at Q 0.7 to 1.0, in the second EQ, after compression and saturation.

Small moves go a long way there because hearing is not flat. The equal-loudness contours standardized in ISO 226 put the ear at its most sensitive around 3 to 4 kHz, so the same level reads louder there than anywhere else in a voice. It is also the region consonants live in, which is what "the words land" means in practice.

Three rules keep this boost out of trouble:

• Bell, not shelf. A high shelf starting at 3 kHz lifts the sibilance band with it, and you have just undone the de-esser.
• Recheck the esses after. A 2 dB lift at 4 kHz raises "sh" and "ch" energy too. One light de-ess pass of 1 to 2 dB after the boost is normal, and the method is in de-esser settings for rap and how to remove sibilance without dulling the vocal.
• If you need more than +3 dB, the problem is not tone. A vocal that stays buried after a 3 dB presence boost is either too quiet, or it is being masked. Raise the fader a decibel, or pull 1 to 2 dB out of the beat in the same region, and the boost you wanted stops being necessary.

Should you boost air above 8 kHz?

Usually yes, as a shelf of 1 to 2 dB starting at 10 to 12 kHz rather than 8. Starting the shelf at 8 kHz puts its knee inside the sibilance band, which for most rap voices runs 5.5 to 8 kHz. Starting it at 10 to 12 keeps most of the lift above that band and still opens the top.

Two things to check before you commit to it. The first is what else you are lifting: room tone in the gaps, headphone bleed, and the noise floor of a budget interface all live up there and all come up with the shelf. Solo the silence between two bars with the shelf engaged and you will hear immediately whether the take can carry it.

The second is that brightness reads as loudness to the machines that measure your record. The K-weighting curve at the heart of ITU-R BS.1770, the same algorithm behind EBU R 128, includes a high-frequency shelf of roughly 4 dB, so a brighter master reads louder before anyone has touched a fader. Spotify measures loudness to that same ITU 1770 standard and normalizes it during playback, so a mix that got its excitement from 12 kHz is turned down to match everything else. Saturation is often the better route to the same perceived top end, and what it does differently is covered in what is a saturator.

Static EQ or dynamic EQ?

Static when the problem is always there. Dynamic when it only shows up on the loud syllables.

The clean way to decide is to listen to a quiet line and a shouted one. A microphone's own peak, a room mode and a consistently boxy space are all present at every level, and a fixed cut handles them with no side effects. A vocal that rings at 900 Hz only when the artist pushes is a resonance in the voice itself, and cutting it statically means the quiet lines are now missing 3 dB they never had a problem with.

The stock route in Logic and Live is a multiband compressor rather than a dynamic EQ. Logic's Multipressor splits the signal into up to four bands and compresses each one on its own (Logic Pro User Guide), and Live's Multiband Dynamics does the same across up to three (Ableton Live audio effect reference). Set the crossovers tight around the problem and that band behaves like a broad dynamic cut. For per-band dynamics on a parametric bell, FabFilter's Pro-Q 4 has a dynamic mode on every band, and Tokyo Dawn Records gives away TDR Nova, a parallel dynamic EQ that will do the job with nothing spent.

There is a third case where neither tool is right: a narrow ring that moves from word to word, following whatever the artist just sang. That is a shifting resonance rather than a fixed peak, and the tool built for it is a dynamic resonance suppressor such as oeksound's soothe3. How that class of processor differs from an EQ is in our resonance suppressor guide.

How does an 808-heavy beat change the EQ?

More than the microphone does. A sustained, distorted 808 is not a sine wave sitting politely at 45 Hz; distortion puts harmonics of that note through the low mids, so a loud 808 has energy in the same 150 to 400 Hz region as the artist's chest. Four adjustments follow:

1. Move the high-pass up and make it steeper. 100 to 120 Hz at 24 dB per octave rather than 80 Hz at 12. The steep slope is what lets you clear the sub while barely touching 150 Hz, which the shallow filter would have reached.
2. Deepen the low-mid cut, slightly. 2 to 3 dB at 200 to 400 Hz instead of 1 to 2, wide. This is the one place where an 808-heavy beat justifies a bigger cut on the vocal.
3. Do the rest on the beat, not the vocal. A dynamic EQ on the beat bus, keyed from the vocal, pulling 1.5 to 3 dB somewhere between 300 Hz and 3 kHz while the vocal is speaking, buys more clarity than any boost on the vocal and costs the beat nothing in the gaps. If you have no sidechain input, a static 1 dB cut on the beat in the presence region does most of it.
4. Raise the compressor's sidechain high-pass to match. 80 Hz is fine on a sparse beat. On this material, 100 to 120 Hz keeps low-frequency spill out of the detector.

Two checks. Mono everything below about 120 Hz, because a wide 808 and a centered vocal fighting over the same octave behaves differently on a club system than on your headphones. And listen on a phone speaker: a small driver reproduces very little of the region the 808 and the chest are arguing about, so the phone tells you whether the words survive and the monitors tell you whether the low end does. You need both answers.

One case where the deeper vocal cut is the wrong fix: if the 808 only plays in the hook, or drops out under the verse, the masking is intermittent, and a static cut on the vocal is paying a constant price for an occasional problem. Use the dynamic version, or automate the cut onto the sections that carry the 808.

What does the tonal EQ after tuning do?

It shapes the voice you are keeping. The first EQ ran on a raw take; by the time you reach the second one the signal has been compressed twice, saturated and tuned, and saturation in particular adds harmonics that change the spectrum you are judging. Shaping before that is shaping something that no longer exists downstream. In the chain article this is stage six, and it sits after pitch correction for the same reason.

The discipline here is restraint: wide bands at Q 0.7 to 1.0, moves of 1 to 2 dB, three bands at most. If the second EQ needs 5 dB somewhere, go back and fix it in the first one, or in the take.

Four regions carry almost all of it, and they are the same four our mastering plugin solves for across a whole song: 50 to 120 Hz, 200 to 400 Hz, 2.5 to 5 kHz and 9 to 15 kHz. MONOLITH measures a long-term average spectrum against a house curve, then halves whatever move that analysis suggests and caps it at 3 to 4 dB, skipping any band with nothing in it. Hand it a finished master and it holds back: in our probe suite, a master reading above 0.8 on its readiness score received under 12 percent of its enhancement and less than 0.3 dB of EQ. Those are song-wide mastering moves rather than vocal moves, but the halving and the cap are the habit worth copying on a vocal too.

One more EQ decision belongs to this stage and lives outside the vocal channel: the reverb and delay returns. High-passing a return at 400 to 500 Hz and low-passing it around 7 kHz removes the fog and the splash without touching the dry voice at all, which is usually the fix when a vocal reads dark in a big space. If the vocal is being tuned as well, the placement argument is in autotune settings for rap.

When is an EQ move wrong?

Nine moves that look reasonable on a chart and are wrong on a rap vocal:

1. Any boost in the first EQ. Cleanup cuts, tone boosts, and the compressor sits between them.
2. A high-pass at 200 Hz on a deep voice. You are on top of the fundamental, not under it.
3. Inverting your sweep settings into a cut. The +10 dB, Q 5 bell was a detector, not a prescription.
4. Any narrow cut deeper than 6 dB on a voice. That is a notch, and notches are for feedback and hum.
5. A wide cut at 1 to 2 kHz to remove honk. Honk is narrow and lower; the wide cut removes the forward edge instead.
6. Cutting 500 Hz on the vocal, the beat bus and the master. Three small cuts sum to one big one.
7. Boosting 3 to 6 kHz to fix a vocal that is simply too quiet. Move the fader first.
8. An air shelf before de-essing, or a de-esser that never gets rechecked after the shelf.
9. Making any of these calls in solo. The vocal is not going to be heard in solo.

The counting rule is a sanity check on its own: two or three cuts in the first EQ, two boosts in the second. A vocal carrying nine active bands has usually been fixed and re-fixed by someone who never stopped to ask whether the take, the microphone position or the room was the problem.

From Our Own Line

We do not sell a channel EQ, so this section is short and it is not a pitch for one. EQ shows up inside two plugins we do make, in ways relevant to the moves above.

TALISMAN, our vocal compressor and saturator, has a sidechain high-pass in its ADVANCED panel that feeds both of its detectors, 80 Hz by default and adjustable from 20 to 500 Hz, which is the filter that decides whether low-end spill or a plosive drives your gain reduction. Its saturation stage also sweeps a high-frequency roll-off from 20 kHz down to 5.5 kHz as you morph from tube toward tape, which is tone control by another name. It is $29.

MONOLITH, our mastering plugin, is the one with a corrective EQ in it: four bands solved from a long-term average spectrum of the whole song, halved and capped at 3 to 4 dB, with empty bands skipped and negative moves allowed so a harsh mix gets pulled down rather than pushed up. It is $29.

Both are VST3 on macOS and Windows, AU on macOS, keyless: no license server, no machine limit. No fake sales. No intro-price games. The price is the price. The rest of the line is on the Ghostnote plugins page.

Frequently Asked Questions

What EQ settings should I use on rap vocals?

In the first EQ, cut only: high-pass at 80 to 100 Hz for most male voices, 1 to 3 dB out of 150 to 500 Hz if the take is boxy, and one narrow cut of 2 to 4 dB at Q 4 to 8 on whatever rings. In the second EQ, after compression and saturation, do the boosting: 1 to 2 dB at 3 to 6 kHz with Q 0.7 to 1.0, and a 1 to 2 dB shelf at 10 to 12 kHz.

Where should I high-pass a rap vocal?

70 to 90 Hz for a deep voice, 100 to 120 Hz for a brighter one, at 12 dB per octave. Remember the filter reaches above its corner: a 12 dB per octave high-pass at 120 Hz is still 1.5 dB down at 150 Hz. High-passing at 200 Hz because a chart said so removes chest from a baritone.

Should EQ go before or after compression on a rap vocal?

Both, doing different jobs. Subtractive EQ goes before, because a compressor's detector reads level and low-mid buildup makes it duck the wrong syllables. Tonal boosts go after, because the detector hears a boost placed in front of the compressor and pulls down the syllables that carry that band.

What frequency makes a rap vocal sound muddy?

Usually 400 to 600 Hz, but mud is masking rather than a fault in the take, so decide it with the beat playing. Try a 2 dB cut at Q 1.0 on the vocal, then try the same cut on the beat bus instead. On most 808-heavy records the beat is the one carrying the excess, and cutting it there keeps the vocal full.

How do I EQ a rap vocal over an 808 beat?

Move the high-pass to 100 to 120 Hz and make it 24 dB per octave, deepen the 200 to 400 Hz cut to 2 to 3 dB, raise the compressor's sidechain high-pass to 100 to 120 Hz, and do the rest on the beat: a dynamic EQ on the beat bus keyed from the vocal, pulling 1.5 to 3 dB between 300 Hz and 3 kHz while the vocal plays.

Should I boost or cut to make a rap vocal sit in the mix?

Cut first, boost second, and keep both small. Two or three cuts of 1 to 4 dB before the compressor and two boosts of 1 to 2 dB after it is a finished vocal EQ. If you need more than 3 dB of presence boost, the vocal is too quiet or it is being masked, and the fader or the beat is the fix.