EasyFader

Guide ยท Dynamics

Parallel Compression: Getting Density Without Losing Punch

A compressor that catches your peaks is also flattening your transients. There is no way around that trade in a single path. Parallel compression is the way around it: compress a copy hard, leave the original alone, and mix the two.

Why Normal Compression Costs You Something

A compressor works by turning the loud parts down. That is the whole mechanism, and it cannot be selective: the same gain reduction applies to the transient and the body of the note, because they arrive at the same time.

So when you compress a snare hard enough to catch 8 dB of gain reduction, the transient is pulled down with everything else. The snare gets smaller. You can make it louder again with makeup gain, but the transient inside it is still smaller relative to the body, and that ratio is what makes a snare sound fat.

This is the fundamental problem, and it is why people who mix rock have spent decades preferring to compress snare and kick less than seems sensible.

What Parallel Compression Does Instead

The signal is split into two paths.

  1. The original path is untouched, or only lightly compressed. It carries the transients, the attack, and the character of the source.
  2. The parallel path is compressed hard, with a high ratio, so it becomes almost completely flat: a dense, sustained, loud version of the source with no peaks left in it.
  3. The two are summed. The result has the original's transient on top of a dense, loud body.

The key thing to understand is that the parallel path is meant to be over-compressed. It is not a subtle effect. A 10:1 ratio with a low threshold and a fast attack turns a snare into a sustained wall of noise, and that wall is what fills the gaps between the hits and makes the part feel bigger.

Why It Sounds Different

The gain reduction is what you hear, not the compression itself. In a normal compressor, all of the gain reduction is applied to the signal you hear, so it is a level change that also changes the balance. In parallel compression, the gain reduction happens on a copy, and what reaches the output is the summation of a compressed and an uncompressed signal.

This is why parallel compression sounds like more, rather than quieter. Adding a compressed copy to an original raises the level of the sustained parts much more than the transient parts, because the compressed copy has no transient to add. The result is density without loss.

It also explains why you cannot simply set a normal compressor to a very high ratio to get the same thing. A 20:1 ratio on a single path means the output is almost entirely gain reduction, and the transient is gone.

How to Set It Up

  1. Duplicate the channel. In most DAWs this is a parallel send to a compressor track, or a duplicated channel with one copy muted.
  2. Set the ratio high. 8:1 to 20:1. The parallel compressor is a levelling tool, not a musical one, so it should be working hard.
  3. Set the threshold low so the compressor is engaging on everything. You want to be seeing steady gain reduction, not intermittent gain reduction.
  4. Set a fast attack, 0.1 to 1 ms. The parallel path should not react to the transient at all, because the transient is already in the original path.
  5. Set the release to suit the material, 50 to 200 ms. On a drum bus you usually want the compressed copy to sustain between hits.
  6. Bring the parallel channel in louder than you expect and adjust by ear.

That last point is where most people go wrong. A parallel channel mixed in quietly sounds like nothing is happening, so it gets mixed in louder, and then it gets mixed in quieter again. Start at parity with the original and pull back from there.

Where It Works

Where It Does Not

It is not a substitute for a good mix. If the parts are fighting for the same space, parallel compression will make them both louder and no clearer. It fixes density, not arrangement.

It is also not a substitute for normal compression. If a vocal is inconsistent in level between takes, that needs a compressor in the main path, because parallel compression cannot correct a level that varies, only add density.

Common Mistakes

Mixing the parallel channel in too quietly. The most common one by a long way. If you cannot hear it, it is not doing anything.

Using a slow attack. The compressed copy then reacts to the transient, which means the transient is present in both paths and gets louder, not preserved.

Forgetting to high-pass the parallel path. The compressed copy carries the same low end as the original, so it doubles the low end of the source. A high-pass at the same frequency you would use on the original keeps the two paths balanced.

Adding it to a bus that is already limited. You end up fighting your own makeup gain, and the loudness creeps up every time you mix the track again.

Questions

What does parallel compression do?

It compresses a copy of the signal hard, then mixes that compressed copy back with the original. The original keeps its transients and dynamics; the compressed copy adds density and level underneath.

Why use parallel compression instead of normal compression?

Because a normal compressor is a trade: you get control over the peaks, but you pay for it with transients. Parallel compression gives you the control on the copy while the original keeps its character.

Should the parallel channel be mixed in louder or quieter?

Louder than you expect. A common mistake is to bring the compressed channel in quietly and hear no effect, then to conclude parallel compression does not work. It should sit close to the level of the original.

What settings should I use?

Hard ratio, 8:1 or more, a fast attack, and a threshold low enough that the compressor is working constantly. It is a different tool to normal compression, and it is supposed to be over-compressed.

Next: sidechain compression โ†’