Vocal mixing: EQ, compression, de-essing, saturation, reverb and delay.

The vocal draws the most attention in any mix, making it the place where processors are most often stacked without a plan. The usual result is a chain inherited from another track: an EQ adding clarity, compression with no known purpose, a de-esser someone inserted out of habit, and reverb leaving the vocal floating.

A useful vocal chain is understood through the problems it solves rather than numbers. This guide follows the usual sequence: correction where appropriate, subtractive EQ, compression, de-essing, saturation, and space through reverb and delay. It finishes with throws, doubles and ad-libs that support rather than mask the lead. A good original take makes everything easier: arrive from vocal recording with healthy levels.

The vocal chain, step by step

The order is not dogma, but it has a logic: first correct what cannot be repaired later, then balance tone, tame dynamics, control the side effects of those dynamics, and finally add colour and space. Each stage affects the next: compress without removing a resonance and the compressor will work on it constantly.

Before chaining anything, set the level: the vocal should enter the mix with margin and without meters shooting up. The level framework is explained in gain staging and dBFS and LUFS meters. Processing an excessively high signal makes every plugin work worse.

EQ: remove before adding

The first move is subtractive. Use a narrow sweep to find troublesome frequencies: boxiness low down, nasal upper mids, isolated whistles. Cut what is unnecessary before raising anything, because every boost costs clarity and space for the other tracks.

A low cut around 80-150 Hz removes background noise and excess body without touching the vocal; a broad, gentle midrange cut can open up the phrase; a high-pass filter below 80-100 Hz is almost always beneficial on sung vocals. Boosts come later: a high shelf for brightness if needed, or targeted presence if the vocal gets lost among instruments.

Order matters here: correcting EQ before compression makes the compressor react to the vocal rather than its defects. A second, finer EQ is often needed afterwards because changed dynamics reveal previously unobtrusive frequencies.

Compression: tame without crushing

Compression makes phrases audible at the same level without the lead disappearing in softer passages. The audible goal is simple: the last word should carry the same weight as the first, and breaths and consonants should not jump out.

As a starting point, iZotope recommends 3 to 7 dB of reduction on the lead vocal; beyond that range it begins sounding squeezed. Threshold determines how much of the phrase is processed, ratio how strongly, attack whether consonant transients pass, and release whether compression breathes or stays stuck. Slow threshold adjustments with moderate makeup gain sound more natural than aggressive control.

If one compressor cannot achieve this without exaggeration, chain two gentle ones: the first tames large peaks, the second settles the whole. If the original vocal is highly dynamic, going back is almost always preferable to squeezing harder.

De-essing: control sibilance

De-essing addresses the side effect of the preceding work: adding clarity with EQ and compressing makes sibilant sounds stand out. In practice, a de-esser is a compressor with a filtered detector: it listens to a specific range and lowers the level only when sibilance appears.

The logical placement is after the clarity EQ and compression, responding to the settled sound. FabFilter documents controls shared by almost all de-essers: threshold, maximum reduction range, and sidechain filters defining the detection window. It places normal vocal sibilants around 8 to 10 kHz, with a search range of 2 to 20 kHz. Single Vocal mode distinguishes sibilant from non-sibilant material; complex material uses the general frequency-range mode.

The listening test is to audition only the detection signal and confirm it contains only sibilants. If you hear the word, the filter is misplaced. Keep the amount low: excessive de-essing makes the vocal murky and consonant-heavy, as though the performer lacked air.

Saturation: body and presence

Saturation adds harmonics and density, making a thin vocal feel integrated rather than placed on top. Use it in small amounts, normally on the lead channel or through a short send.

Judge the result rather than circuit type: a saturated preamp adds warmth, tape softens peaks, and a more aggressive mode adds grit. Less is more in every case: if saturation is audible as distortion, it is already too high. It also works well on doubles, separating them from the lead without raising their volume.

Reverb and delay: depth without clutter

Reverb and delay give the vocal a place. Reverb softens and adds size; delay adds movement and repeats information supporting long phrases. Both share the same risks: excessive tails compete with intelligibility, and low-frequency content drowns the voice.

The usual controls are pre-delay, separating voice from tail and keeping it forward; removing low and high frequencies from the reverb so it does not compete with body or consonants; and amount, almost always lower than it seems in isolation. Instead of inserting space on the channel, usually use a send, sharing it among vocals and automating it without touching the lead track.

Throws, doubles and ad-libs

Throws. Delay triggered on one word or phrase, usually at a section’s end, and cut before the next entrance. It adds an arrangement detail a constant echo bed cannot, automated simply by opening the send at the exact point.

Doubles. A second performance of the same text, thickening the chorus and adding width. They need their own space: lower level, some stereo spread and sometimes different processing from the lead. Sounding identical to the lead at identical level produces combing rather than body.

Ad-libs. Supporting phrases, exclamations and flourishes over the top. Treat them as separate elements: more processing, more space and clear gaps so they do not overlap the verse. All three resources share one rule: support rather than compete. If the ear does not know what to listen to first, there is too much.

Next step

Once the vocal is assembled and mixed, bring it to the digital stage: DJ acapellas with stems, de-noise, de-reverb, chops and stutters.