Parallel Compression on Vocals: Loud and Natural at the Same Time
Vocal mixing · 4 min read
Send the vocal to a bus, compress that bus hard (ratio 10:1, fast attack, 10–15 dB of gain reduction), then blend it back under the dry vocal until the quiet words come forward. The dry track keeps the dynamics; the crushed copy adds the weight.
Blend to taste
10–15 dB
Gain reduction on the parallel bus · blended 20–35%
The 60-second answer
Create a send from the vocal to a new bus. On the bus: ratio 10:1, attack 1–5 ms, release fast, threshold low enough for 10–15 dB of gain reduction. Bring the fader up from silence until the vocal feels solid, then back it off 2 dB.
Why this works when normal compression does not
Regular compression is a trade: control now, life later. Parallel compression cheats it. The dry signal keeps every bit of the original dynamic range, and the crushed copy underneath fills in only the parts that were too quiet to hear. The loud words barely change; the whispered ones get lifted into the track. That is why it sounds louder without sounding compressed.
The one thing that ruins it: latency
If your compressor has any lookahead or oversampling, the parallel copy arrives a hair later than the dry track, and the two combine into a comb filter that eats the low mids. Symptom: the vocal gets thinner when you blend the bus in, not fatter. Turn on delay compensation, or turn off oversampling on that plugin.
The setup, step by step
| # | Do this |
|---|---|
| 01 |
Make a send, not a duplicate Post-fader send from the vocal to a new stereo bus called PARALLEL. Post-fader means your automation still works. Send: post-fader, -inf to start |
| 02 |
Put one compressor on the bus and go extreme This is the one place where ugly settings are correct. Fast attack, fast release, high ratio. Ratio 10:1, attack 1–5 ms, release 50 ms |
| 03 |
Pull the threshold down until it is crushing 10–15 dB of gain reduction. It should sound bad in solo. That is the point. GR 10–15 dB |
| 04 |
High-pass the parallel bus at 150–200 Hz The crushed copy brings up mic proximity and room noise. Cutting the lows means you are only adding body, not mud. HPF 150–200 Hz |
| 05 |
Bring the fader up from silence Never start at 0 and pull down. Start at -inf, raise until the quiet syllables step forward, then lower it 2 dB. Blend: 20–35% |
| 06 |
Check in the full mix, then check on your phone Parallel compression is the effect most likely to sound great in solo and crowded in context. |
The trick: mute the dry vocal for two seconds
With the blend set, solo the parallel bus alone. You should hear the breaths, the room tone, the mouth noise — all the stuff that lives 20 dB below the words. If you hear mostly the words, your threshold is too high and you are just adding a second copy of the vocal, which is what makes parallel compression sound like a doubled, phasey mess. The parallel bus is supposed to be the sound of everything that was too quiet. Set it there, and it disappears into the mix instead of fighting the lead.
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
- Blending too much. Above roughly 40% you stop hearing dynamics at all. The vocal reads as loud but flat, and no amount of EQ fixes it.
- Using a duplicate track instead of a send. Any fader or automation move on the lead then desyncs the two, and the balance shifts through the song.
- Skipping the high-pass. Crushed low mids stack with the beat and make the whole mix feel small.
- Adding parallel compression to a vocal that has no dynamics left. If your serial compressor already did 8 dB, there is nothing quiet left to lift.
Keep going
FAQ
Is parallel compression the same as New York compression?
Yes. Same technique, different name, borrowed from drum mixing in New York studios in the 80s.
Should parallel compression go before or after EQ?
After your corrective EQ, so the crushed copy is not amplifying a resonance you already decided to remove.
Can I use a mix knob instead of a bus?
Yes, if the plugin has one and compensates delay properly. A dedicated bus gives you EQ control over the parallel path, which is where most of the benefit is.
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 →