2. Two-stage compression
Why two gentle compressors sound better than one working hard, what values to start with, and how to keep the vocal from getting crushed.
Without compression, the soft parts get lost under the guitars and the loud ones jump out. A rock or metal vocal almost always takes quite a lot of compression, and the way to keep it from showing is to split it up.
Two compressors instead of one
If a single compressor has to solve everything, it works hard and you can hear it. With two, each one does little.
Compressor 1: for the peaks
Compressor 2: to even things out
- Attack: slower, 10 to 30 ms, to let the consonants through.
- Release: medium, 100 to 200 ms.
- Ratio: 2:1 or 3:1.
- Threshold: until you get 2 to 3 dB of nearly constant reduction.
What to listen for
- Too little: some words disappear under the band.
- Right: you follow the lyrics effortlessly and the voice feels stuck to the front.
- Too much: breaths and room rise, sibilance gets exaggerated, and the voice sounds flat or pumps.
Always compare at the same volume. Set the output gain so the bypass doesn’t change the level: otherwise, the louder one will always seem better.
Try it
Load your vocal, bring the threshold to about 3 dB of reduction, and alternate between fast and slow attack. With fast, the consonants lose their edge; with slow, they come back.
Playground · Compressor
- •▶ Play — plays the audio in a loop.
- •No compression / With compression — switch between the sound with and without compression to compare, at the same volume. The segment with the lit border is the one you're hearing (“A/B” comparison).
- •🎸 Bass — a synthesized riff with dynamics: ideal for hearing compression.
- •📢 Pink noise — an even sound across the whole spectrum: ideal for “seeing” the EQ.
- •🎵 Load your audio — your own file (MP3, WAV, OGG, FLAC, M4A…); it's processed locally in your browser and isn't uploaded anywhere.
See how the output line (orange) stays more even than the input line (gray): the compressor brings loud and soft closer together. The more you compress, the flatter it gets.
The compressor evens out the volume (it lowers the loudest parts). Move each control and watch the GR bar above:
A recipe to hear it clearly: source 🎸 Bass, ratio 4:1, and lower the threshold until you see 3–6 dB on the GR bar. Turning the compression on and off you'll notice the bass gets more even and “glued”.
Compression is what lets the vocal be understood from start to finish over a dense band. It gets refined later, but it has to be there before you decide the balance and the color EQ.
Growled and screamed vocals
A screamed vocal already has little dynamic range and a lot of energy in the mids, and it usually handles more compression, with a more obvious sound: there it can be part of the character. Use the same structure, with more reduction on the first compressor, and process them on a separate track from the clean vocals.
Common mistakes
What NOT to chase yet
Quick path: compression
Quick path
- Compressor 1: fast attack, ratio 4:1, 3 to 5 dB only on the loudest parts.
- Compressor 2: slow attack, ratio 2:1 or 3:1, 2 to 3 dB nearly constant.
- Match the output volume and compare with the bypass.
- Listen for whether you understand the lyrics with the band playing.
- If breaths or room rise too much, ease off.
Expected result
Expected result
A vocal that’s even and up front, where you follow the lyrics without effort and without noticing a compressor. The side effect is that sibilance is now more audible. We deal with that in the next section.