Why compress a copy instead of the original?

Because the original then keeps its transients intact. The compressed layer is extremely dense and lifts quiet detail, while the dry copy carries the attack. Summed together, you hear both. Compressing the original directly would give the density without preserving the punch, which is a different sound.

What blend percentage should I start with?

The default of 40 percent is a sensible middle. Drums usually take more, somewhere between 30 and 50 percent, while vocals want less because the density competes with intelligibility. Anything above about 50 percent makes the compressed layer the dominant sound rather than a support beneath the original.

Why is the internal compressor so aggressive?

Because its job is to be a dense layer rather than a transparent processor. A threshold of minus 28 dBFS with an 8 to 1 ratio and 6 decibels of make-up gain makes it work hard on purpose, so that its quiet detail sits high enough to be heard when blended under the original. Tuning it to be gentle would defeat the technique.

Will the output be louder than the input?

Yes, and that is expected, since two layers are being summed. There is no limiter in the chain, so a high blend on already hot material can exceed full scale. Check the level after the pass and normalise or limit as needed rather than assuming the result is safe.

Does it work on spoken word?

It can, with restraint. A small blend adds body to a dry voice without making it sound compressed, which is useful in podcast work. The catch is that the compressed layer also raises the noise floor between words, so on a noisy recording the effect can be counterproductive.

Parallel Compressor

You Might Also Need

Dynamic Range Compressor

Professional-grade audio compression

Audio Compressor

Dynamic range compression for audio

Sidechain Compressor

Ducking and pumping effects

Multiband Compressor

Compress different frequency bands separately

Parallel Compressor blends a heavily compressed copy underneath the untouched original, so detail and sustain come up while the punch of the dry signal stays. It is the classic technique for drums, and it works on voices and bus mixes too.

What is Parallel Compressor?

Drop a file, set how much of the compressed layer to blend in and how aggressive that layer should be, then read the report. The footer states the ratio and threshold used for the layer and the blend percentage, so the two halves of the decision are both on record.

The controls:

A blend control from 0 to 100 percent in steps of 5, defaulting to 40, with the remainder staying dry.
A compressor threshold from minus 48 to minus 12 dBFS in single decibel steps, defaulting to minus 28.
A compressor ratio from 2 to 20 in single steps, defaulting to 8.
An internal compressed layer fixed at a 3 decibel knee, plus 6 decibels of make-up gain, 3 millisecond attack and 80 millisecond release.
A report line naming the ratio, threshold and blend, plus the dry share as a percentage.
An output WAV named after the source with a parallel suffix, limited to two channels.

Parallel compression works because the two halves do different jobs. The compressed layer is deliberately aggressive, with a low threshold and a high ratio, so quiet detail behind the main hits is lifted heavily in relative terms and sustains longer. Blending it under the untouched original at a partial level means the transient information that a heavy compression pass would have softened is still present in full, because the dry signal carries it unchanged. The result is denser without being flatter, which is why the approach is associated with drums, where both attack and body matter. Blend is the control that decides the character: at a low percentage the compressed layer adds a little support beneath the original, while at 50 percent or more the density becomes the dominant quality. The compressor in the layer is fixed at its own attack, release and make-up settings on purpose, because the only thing worth tuning is how hard it works and how much of it you hear.

What the blend cannot do:

It raises the level as well as the density, so the sum of two layers can exceed full scale and there is no limiter to stop it.
A high blend percentage on already compressed material simply adds more compression and little else.
The dry layer is untouched, so any problem in the original, including noise, comes through at full level.
The internal compressor's attack, release and knee are fixed, so the layer's character cannot be tuned beyond its threshold and ratio.
There is no metering, so the only guide to how much the layer is working is the blend percentage you set.
It works on the whole file, so it cannot apply the technique to one element inside a mix.

Use it when the goal is density rather than control: a drum bus that needs more body, a vocal that needs to sit forward without sounding squashed, or a mix that needs weight without losing its punch. Start at the default blend, listen for whether the result is denser in the right way, and remember that the level will have risen, so normalising afterwards is the usual last step.

How to use Parallel Compressor

Drop an audio file onto the panel; the tool accepts up to two channels.
Set how hard the internal compressor works by choosing a threshold and a ratio for the compressed layer.
Blend the compressed layer in, starting from the default of 40 percent and adjusting from there.
Read the reported blend and settings, then download the parallel processed WAV.

When to use Parallel Compressor vs related tools

Parallel Compressor is the tool for adding density while keeping transients, which makes it a natural fit for drums and for vocals that need presence. Dynamic Range Compressor is the direct route when you want the compression itself to be audible in the main path, and Audio Compressor exposes the raw controls. When the aim is to keep a music bed out of the way of a voice rather than to add density, Sidechain Compressor is the tool built for that instead.

Privacy & Security

This tool runs entirely in your browser — no data ever leaves your device. There is no server round-trip, no upload, no logging, and no account required. Your input is processed locally using client-side JavaScript and is never stored, transmitted, or accessible to anyone else. When you close the tab, everything disappears.