A vocal is the hardest element in most mixes to make sit well. It is the one part of the arrangement the listener usually cares about most, it is recorded in the least controlled conditions, and it is the part most likely to have a performer moving relative to the microphone throughout the take.

The chain you build around it should reflect that. Subtractive processing, which removes what is not there, should come before additive processing, which adds what is missing, and any plugin that is not fixing a specific problem you have identified is adding cost without benefit.

The order that works

Filtering comes first, because everything downstream benefits from a signal without inaudible rumble. Then corrective EQ, which addresses problems in the recording itself such as a boxy resonance or a proximity effect that was too strong. Then compression, which controls the dynamic range that the performance created. Then a de-esser, which addresses sibilance that the compression itself may have made worse.

Additive processing comes last. Reverb, delay, and any saturation or excitement that is presentational rather than corrective. These belong on sends in most workflows, so that the dry vocal remains a clean, controllable signal and the effects are added around it.

Making every plugin earn its place

A common habit is to load a chain from a template and leave it there. This produces two predictable problems. Plugins set for one voice sit awkwardly on another, and plugins doing nothing but costing CPU and headroom create the illusion of a more considered mix than actually exists.

The discipline that works is to bypass each plugin in turn and note what changes. If the answer is nothing audible, remove the plugin. A vocal chain should be as short as the recording allows, and many excellent mixes use three or four plugins on the vocal and nothing else.

Compression specifics

Vocal compression is usually about control and consistency rather than about loudness. The goal is to even out the phrase so quiet words are not inaudible and loud words are not uncontrolled, and then use level automation for the parts that automation handles better.

A ratio of 3:1 to 4:1 with a moderate attack is a reasonable starting point. Faster attacks sound more processed on plosives and consonants because the compressor reacts to the transient rather than to the sustained pitch. Slower attacks let the transient through and compress the body, which usually sounds more natural but provides less peak control.

Two compressors in series is occasionally right, and occasionally it is a way of avoiding a decision about the first one. If you are reaching for two, the usual reason is that the first handles peak control and the second handles the sustained body, and that division should be deliberate.

Building the chain from the recording

1

Start with the raw take, level-matched and with all effects bypassed. Judge the performance, not the processing. Compression applied over a poorly performed take hides the problem rather than solving it.

2

Filter, then correct the problems you can hear. Proximity effect, a boxy resonance, mud. Fixing these before compression gives the compressor a more consistent signal to work with.

3

Compress for consistency, then automate for phrasing. The compressor evens things out. The automation makes the performance read. Doing both jobs in one plugin usually means neither is done well.

4

Add sends last, and check the vocal in mono. Sibilance and masking problems are much easier to hear in mono, and a vocal that disappears in translation was never really present.

Vocal chain order and purpose

StageTypeJob
High-pass filterSubtractiveRemove rumble and proximity build-up
Corrective EQSubtractiveFix problems in the recording
CompressorDynamicControl range for consistency
Second compressorDynamicOnly if peaks and body need different control
De-esserSubtractiveControl sibilance, often post-compression
Tonal EQAdditivePresence and colour, used sparingly
Reverb and delayAdditiveOn sends, never inserted

Retuning a processed vocal

A retune changes the frequencies in the vocal, which interacts with every stage of the chain. A high-pass at 100Hz is unaffected. A presence boost aimed at a particular range may now be landing differently, and a de-esser set to a specific frequency will be targeting a sibilance peak that has moved with the rest of the signal.

The reliable approach is to re-bypass and recheck the chain after retuning with 432Hz MASTER. Narrow corrective settings are the ones that will have drifted, and they are also the ones that sound worst when they are wrong, because they are acting precisely where the problem is no longer located.

With RASTA 528Hz, where the harmonic content changes rather than sliding, the whole chain deserves a proper recheck rather than an adjustment.

The bottom line

Subtractive first, additive last, and every plugin justified by a problem you can actually hear. A short chain built from the recording will beat a long one copied from a template.

Frequently asked questions

What order should vocal plugins go in?

Filtering, then corrective EQ for problems in the recording, then compression for consistency, then a de-esser for sibilance that compression may have amplified. Reverb, delay and presentational processing go last, normally on sends rather than inserted.

How many compressors should I use on a vocal?

Usually one. Two are occasionally right when you want fast peak control followed by slower control of the sustained body, and that division should be deliberate. Two compressors with similar settings usually means the first one was not set properly.

Should I EQ before or after compression?

Corrective EQ belongs before compression, so the compressor receives a more consistent signal. Tonal or additive EQ can go after, once the dynamics are settled. Moving EQ after compression is what makes sibilance worse, because compression boosts the sibilant peaks along with everything else.

Does retuning affect my vocal chain settings?

Yes. Every frequency-dependent plugin in the chain is now targeting a location the signal has moved from. High-pass filters placed low are unaffected, but presence boosts and de-esser settings will have drifted and should be rechecked after retuning.

More from the atixUniverse blog