Why isn't my vocal audible even when I turn it up?

Because loudness and intelligibility aren't the same thing. If other elements occupy the 1 to 4 kHz zone — guitars, synths, cymbals — the vocal arrives louder but just as buried. What uncovers it is making room there by lowering that zone in whatever is masking it, not raising the fader.

What order do the plugins go in?

The usual order is: high-pass filter, resonance cuts, de-esser if needed, compressor, colour EQ and finally effects on sends. The logic is to clean before compressing — so the compressor doesn't react to what you were about to remove — and colour afterwards, once the dynamics are stable.

How much compression does a vocal need?

More than almost any other track, because the distance between a soft and a loud syllable is enormous. It's common to split it across two gentle stages rather than one hard one: 3 or 4 dB in each draws far less attention than 8 dB at once and controls just as well.

What do I do about the esses?

First check whether you're creating them: a high-frequency boost or aggressive compression turns ordinary sibilance into a problem. If they still bother you, a de-esser working between 5 and 9 kHz solves most cases. Automating the volume of the specific esses is more work but sounds better.

Reverb or delay on the vocal?

Delay if you want depth without losing clarity, because the repeats fill gaps rather than everything all the time. Reverb if you want to place it in a space. Combining both with a small amount of each usually works better than loading up one alone.

Is it better to automate or to compress more?

Automate first, compress second. An automation pass that evens out phrases and sections leaves the compressor only the fast work — syllable to syllable — which is what it's good at. Compressing without automating forces the compressor to fix long-range imbalances, and that's where it starts sounding squashed.

Mixing12 min read

How to mix vocals so they sit forward and stay there

Chain order, where to EQ, how much to compress, what to do about sibilance, and how to make room for the vocal without touching the fader.

The problem is almost never the vocal

«The vocal isn't audible» is the most repeated complaint in any mix, and the automatic reaction is to turn it up. That's nearly always the wrong answer, because the symptom isn't about volume: it's about room. A voice's intelligibility lives mostly between 1 and 4 kHz, and that strip is exactly where electric guitars, pad synths, the snare and much of the cymbals also keep their presence.

When five elements fill that zone the vocal competes on equal terms and loses, because it has the least continuous energy. Turn it up and all you get is a louder mix with the same problem — and you eat your master headroom on the way. What works is deciding that the vocal owns that strip, and relieving it in whatever masks it.

The chain, in order and for a reason

High-pass between 80 and 120 Hz depending on the voice: clears room rumble, footfall and plosives without touching the body.
Narrow resonance cuts: nearly any voice recorded in an ordinary room has one or two annoying peaks between 200 and 600 Hz.
De-esser, if needed, BEFORE the compressor: that way the compressor doesn't react to esses you were going to tame anyway.
Compressor, or two in series: the first catches peaks, the second levels the body.
Colour EQ at the end: presence between 2 and 5 kHz, air with a shelf above 10 kHz.
Reverb and delay always on sends, never as inserts: that way you control the amount without touching the dry vocal.

Automate before you compress

A vocal performance has two very different kinds of imbalance: the long one — a verse weaker than the chorus, a phrase that fades at the end — and the short one, between syllables. A compressor is only good at the second. Ask it to fix both and it has to work so hard that you hear it working.

Which is why the order that works best is the counter-intuitive one: first a rough automation pass, lifting the phrases that drop and pulling down the ones that jump, and only then the compressor. It costs twenty minutes and saves three or four dB of compression — exactly the difference between a vocal that breathes and one that sounds like a wall.

Placing it in space without losing it

A completely dry vocal sits stuck to the listener and gives away the room it was recorded in. A vocal with too much reverb drifts to the back and stops being intelligible. You don't find the middle by pushing the send until it «sounds good», but by deciding how far away you want it and using pre-delay to keep the consonants clear.

A pre-delay of 20 to 40 ms lets each word's attack through before the tail arrives, and with that you can use considerably more reverb without losing intelligibility. How it works in full — and why you should EQ the effect rather than the source — is in the reverb and delay guide.

And if the problem is that you can't hear where the masking is happening, that trains: the EQ Surgeon drills that exact skill, and the map of the spectrum tells you where to look while you build it.

Frequently asked questions

Why isn't my vocal audible even when I turn it up?: Because loudness and intelligibility aren't the same thing. If other elements occupy the 1 to 4 kHz zone — guitars, synths, cymbals — the vocal arrives louder but just as buried. What uncovers it is making room there by lowering that zone in whatever is masking it, not raising the fader.
What order do the plugins go in?: The usual order is: high-pass filter, resonance cuts, de-esser if needed, compressor, colour EQ and finally effects on sends. The logic is to clean before compressing — so the compressor doesn't react to what you were about to remove — and colour afterwards, once the dynamics are stable.
How much compression does a vocal need?: More than almost any other track, because the distance between a soft and a loud syllable is enormous. It's common to split it across two gentle stages rather than one hard one: 3 or 4 dB in each draws far less attention than 8 dB at once and controls just as well.
What do I do about the esses?: First check whether you're creating them: a high-frequency boost or aggressive compression turns ordinary sibilance into a problem. If they still bother you, a de-esser working between 5 and 9 kHz solves most cases. Automating the volume of the specific esses is more work but sounds better.
Reverb or delay on the vocal?: Delay if you want depth without losing clarity, because the repeats fill gaps rather than everything all the time. Reverb if you want to place it in a space. Combining both with a small amount of each usually works better than loading up one alone.
Is it better to automate or to compress more?: Automate first, compress second. An automation pass that evens out phrases and sections leaves the compressor only the fast work — syllable to syllable — which is what it's good at. Compressing without automating forces the compressor to fix long-range imbalances, and that's where it starts sounding squashed.