Serial Compression on Vocals: Two Compressors Beat One
Vocal mixing · 4 min read
Serial compression means two compressors in a row, each doing a small amount. Put a slow opto-style compressor first for 2–4 dB of level control, then a fast FET-style compressor for 2–3 dB of peak catching. The trick is that the two attack times must be far apart, or you have just built one compressor with a steeper ratio.
Total gain reduction
3 + 3 dB
Opto at 10 ms attack · FET at 0.3–0.8 ms
The 60-second answer
Compressor 1: opto style, ratio 3:1, attack near 10 ms, program-dependent release, 2–4 dB of gain reduction on the loud phrases. Compressor 2: FET style, ratio 4:1, attack 0.3–0.8 ms, release 100–200 ms, 2–3 dB on the spikes only. Level-match and A/B against bypass.
Why two small moves beat one big one
Every dB of gain reduction costs you a little transient detail and a little of the level swing that makes a voice sound alive. That cost is not linear. Two dB is free. Four dB you can hear if you look for it. Eight dB on one plugin sounds like a plugin, whichever one you bought. Split the same total across two units and each stays inside the range where it is invisible.
What each stage is actually for
Stage one is a leveller. Its job is to make the verse sit in a narrow window so the fader stops mattering. Opto circuits do this well because the release is program dependent: roughly half the gain reduction recovers in about 60 ms and the rest takes half a second or more, which reads as a hand on a fader. Stage two is a peak catcher, for the handful of consonants and belted notes still poking 4 dB above everything. That needs a sub-millisecond attack, and a sub-millisecond attack is the wrong thing to put across a whole performance.
The two-stage setup, step by step
| # | Do this |
|---|---|
| 01 |
Ride the clip gain first Get the loudest and quietest words inside a 3 dB window before any compressor. Serial compression is for the last few dB, not the first twelve. Clip gain: +/-3 dB window |
| 02 |
Compressor 1: slow, gentle, always working Opto or vari-mu style. Pull the threshold down until you see 2–4 dB on the loud phrases and near zero in the quiet ones. Ratio 3:1, attack 10 ms, GR 2–4 dB |
| 03 |
Makeup gain back to unity Match output to input by ear so compressor 2 sees the level it expects. Skip this and your second threshold is guesswork. Makeup: level-matched |
| 04 |
Compressor 2: fast, high ratio, barely moving FET style, attack as fast as it goes on most 1176-type plugins. Threshold high enough that it only catches spikes. Ratio 4:1, attack 0.3–0.8 ms, GR 2–3 dB |
| 05 |
Watch both meters at once Stage one should move slowly and almost continuously. Stage two should flick six to ten times a verse. If they move together, the settings are wrong. GR meters: not in sync |
| 06 |
Level-match, bypass both, then decide The compressed version should sound the same size and more consistent, not louder. Cap the total at 6–7 dB. Total GR: 6–7 dB max |
The trick: keep the two attack times a factor of 10 apart
If both compressors have similar attack times, they react to the same syllable at the same moment, and 3 dB plus 3 dB becomes 6 dB of gain reduction on one event. That is not serial compression, that is one compressor with a steeper ratio, and it sounds like it. So set them deliberately apart: one at 0.3–0.8 ms, the other at 10–30 ms, at least a factor of ten between them. Do the same with release — 100–200 ms on the fast unit, 400 ms to a second on the slow one. Now the fast unit handles events a couple of milliseconds long and the slow unit handles events a couple of seconds long, and neither ever sees what the other dealt with. Test it by watching both gain reduction meters through a busy verse: if the needles move in lockstep, widen the gap until they clearly disagree.
Stop guessing the chain.
The Vocal Chain Bible is 88 real vocal chains from Grammy-winning engineers — every plugin, every setting, in order, with free alternatives for each one. 5+ years, 100+ interviews, one PDF.
Get the Vocal Chain Bible — $48 →Mistakes that kill it
- Same settings twice. Two instances with the same attack, release and ratio is one stage doing double the work with double the artefacts.
- Chasing a total number. Nobody hears 6 dB of gain reduction. They hear whether the quiet words are audible. Set by result, not by meter.
- No makeup gain between the two. The second threshold only means something relative to the level arriving at it. Reset to unity every time.
- Both compressors after the saturation. Saturation already flattens peaks, so the fast compressor has nothing left to catch.
Once both stages are set, the vocal fader stops moving during playback and you can hear what the EQ is doing. Stable level means every decision after it is about tone.
Keep going
FAQ
Should the opto or the FET compressor go first?
Either works, but slow-then-fast is easier to control: the slow unit levels the performance and the fast one cleans up what escaped. Fast-then-slow is more aggressive and common on rap vocals.
How much total gain reduction is right?
Around 5–7 dB split roughly evenly. If you need more, ride the clip gain first or record the line again.
Can I use the same plugin twice?
Yes, as long as you change attack, release and ratio so the two instances work on different time scales.
Who's behind this
Ivan Meshcheriakov · IvanFromRnD
5+ years ago I moved to LA with zero connections and started building from nothing — 14-hour days, meeting people and interning at studios. It took me 5+ years to meet and interview 100+ Grammy-winning engineers — then I turned it all into the Vocal Chain Bible.
Read my full story →