The Science Of Art: Parallel Compression

Following on from the last post on Serial Compression (back in 1872), I thought it would be good to post a guide on Parallel Compression. I was actually writing a post on mix bus (aka master bus) compression and thought it’d be a good idea to include this topic first as I might refer to it a bit in the next post. Click here for a handy overview of all 5 previous parts of the series on compression.

Random picture of some DBX compressors @ KONK Studios (London) so that you don’t glaze over from looking at all this text

Let’s get right into it: parallel compression is when you blend a compressed version of a track with the original track. That’s literally all there is to it, don’t let anybody scare you into thinking it’s more difficult or complex than that.

It’s sometimes called ‘New York style’ compression by people who want to make it seem like they are more elite than you and have top secret references and knowledge that you simply don’t understand.

It allows you to use to extreme compression settings without completely destroying the transients (aka the peaky out-y bits) of your source audio - that’s why it’s often used on drums, you can use it to bring out the room sound of the drums and the sustain of the cymbals - without absolutely trashing those spiky transients, which help give the ‘punch’. But you can use it on any source you’d like, of course (including the mix bus…). Simple, but cool.

Wet/Dry Blend Knob

Nowadays, most plugins have a mix knob or a way to blend the dry/wet levels (dry being the uncompressed or unaffected signal, and wet being the compressed signal), so you can achieve parallel compression just by tweaking this mix knob - you’re probably already doing it.

If you’re using a hardware compressor alongside your computer, you can use an I/O insert plugin to insert the compressor/FX, then there’s usually a mix slider on the plugin. Just be sure to hit the ‘ping’ button (other DAWs will have something similar), which will compensate for the latency introduced by the round trip out of your sound card and back.

There are a couple of other ways to achieve parallels compression that gives you a lot more flexibility: by ‘multing', and using sends.

Multing

Another way to achieve the same effect is to simply copy your track, compress the crap out of it, turn it down, then bring it up to blend it with the unaffected track. This is also sometimes called ‘multing' - e.g. splitting a guitar into compressed and uncompressed by multing the same guitar recording and sending one to a compressor blending the two sounds. And you don’t need to only use compression, you can EQ it, add FX etc, then blend it in with the original track.

Some people even ‘mult’ a vocal and split it into three using EQ - low end, mids and highs, treating them all differently and blending them all back together - but those people are usually mad.

This technique is cool as you can automate and add a bunch of EQ and other effects to the ‘multed’ track.

Send It / Go Full Send / Other Word Play

You can also use a ‘send’ to send the audio to a bus, and then compress that bus - you’ve probably already done this with reverb. For example, you could make a ‘smash bus’, that you send the kick and snare to, and blend that in a tiny bit with your original kick and snare. Or just the drum room mics for a section of the song. You can even gate the input, if you want. Go wild, it’s your party.

It’s a reference to Lost In Space, incase you were wondering

DANGER WILL ROBINSON

One thing to watch out for when using the last two methods: don’t get tricked into thinking it sounds better just because you’re added more ‘stuff’ to the original signal. Obviously if you take the original sound, duplicate it, compress the crap out of it and then blend it in with the original, it will sound louder because more is more. But you’re too smart to fall for that, I believe in you.

Adding a low or high pass filter can make a huge difference to the sound of the compressed track and can enable you to go way more extreme with the compression (or distortion, at this point). Remember that you can EQ the signal going into the parallel compressor, OR after the parallel compressor. Try chopping a load of low end off before it hits the compressor and compress really hard, then blend back in with the original…

Don’t feel the need to use loads of the compressed/crushed signal.

Finally, if you’re trying to use a tape machine or tape machine plugin to get tape compression, parallel compression might not work due to wow and flutter (when using a plugin, you might be able to turn this off) - but it will probably sound like a cool phase/flanger effect!

When Do I Really Use It Though

I tend to use it more with extreme distortion levels of compression (or just distortion/saturation full stop) or when the compressor or effect adds a bit of ‘colour’ - but I don’t want too much. I automate the effect in/out of sections a LOT.

In terms of pure compression, I use it when I ‘kinda like’ what the compressor is doing but know that deep down it’s detracting from the original source. So I’ll try to turn down the mix knob and see if I can have the best of both worlds.

Try It Backwards

Something I often try towards the end of a mix: if things are sounding a bit ‘squashed’ or I want more of e.g. a drum element, I’ll send some of the ‘dry’ drum - e.g. a kick and snare direct track - directly to a the master bus, bypassing the drum bus compressor. Because it’s hitting the mix bus and bypassing the drum bus compression and effects, it gives a completely different sound that can be really cool.

If you think about it, it’s a bit like the inverse of parallel compression - I’m adding more of the original signal on top of the compressed version. You can do this e.g. for just the choruses when everything kicks off once the guitars kick in and the snare gets a bit lost - you might need the snare to poke out more when the mix bus compressor is clamping down on the whole mix. You could automate up the ‘extra’ snare just on the chorus for a bit of extra impact.

Alright, this is getting a bit complicated now.

So anyway: parallel compression is when you blend a bit a compressed signal with the original signal and that’s it, go experiment and have fun with it.

Next up, mix bus compression.

With love,

Ed ❤️🎚🎛

The Science Of Art: Serial Compression

Ahoy - this time we’re talking about serial compression - exactly what’s going on, and why you shouldn’t be too scared of using multiple compressors in series.

Serial compression can be very straightforward: it’s simply 2 or more compressors in series, one after another, with each compressor carrying a little (or a lot) of the workload. It can be used on individual tracks, on the master, or on any busses - in fact it would be rare that any song released these days wasn’t run through a few compressors in series, during mastering.

Essentially, the reason to be aware of/understand serial compression is because you can achieve results with it that you just can't get with a single compressor.

For example, when mixing individual tracks, I’ll often use a fast-reacting compressor with a higher ratio first to catch the higher transient peaks and increase the ‘short-term consistency’ of the audio. Then I’ll follow it up with a slower compressor at a lower ratio to deal with more long-term level variations. This way, the second compressor has a smoother, more balanced signal to work on and won't ‘pump’ or ‘breathe’ due to a sudden burst of energy on signals with occasional loud peaks. I find it really helps with dynamic vocals and inconsistent bass guitar. The classic chain for this kind of thing is an 1176 into an LA2A.

This can often result in a more transparent, natural sounding compression, as well as reduce potential negative artefacts caused by heavy compression, such as treble loss, and/or plugin aliasing - some plugins can sound okay when doing gentle compression but kind of ‘fall apart’ when being pushed. You can also benefit from the colouration or ‘character’ that both compressors will impart on the sound.

However, using two or more compressors means it’s quite easy to over-compress…

Multiplicative Ratios!

Consider a vocal that’s been recorded through a hardware compressor using a 12:1 ratio, followed by two more compressors used in serial during mixing, with a ratio of 8:1 and 2:1. Now let’s say there’s also an SSL bus compressor on the master bus applying a 4:1 ratio (which is a totally reasonable scenario for a mix I might do). What’s the full potential compression ratio that this vocal track goes through?

The answer is…

768:1 (!)

The reason: Compression ratios are multiplicative when over the threshold: 12 x 8 x 2 x 4 = 768

Given that any ratio above 10:1 is considered limiting, 768:1 seems a ‘little extreme’, and that’s before mastering with more compression and limiting. Even with no compressor used during tracking, there’s still potentially a 64:1 compression ratio being applied to the vocal. So yeah… it’s easy to over-compress using serial compression.

To break it down and really help us understand it: imagine two compressors in series, both set at a threshold of -10dB, and both with a ratio of 4:1. Theoretically, this should result in a potential 16:1 ratio using serial compression.

So, if we imagine the audio peaks at -2dB (8dB over the threshold) enters the compressor:

    •    After 4:1 compression the signal is now only 2dB above threshold
    •    Compress again with 4:1 ratio and signal only goes 0.5dB above threshold

So yeah, going from 8dB (initial signal above threshold) to 0.5 gives a 16:1 ratio (because 8 divided by 0.5 is 16).

Real World Usage

But these figures don’t tell the whole story as, in reality, serial compression is rarely (if ever) done by setting the thresholds of the two compressors identically, so the aggregate ratio of compression is much more ‘fluid’. Additionally, both compressors probably wouldn’t have the same attack and release times, the same knee, and so on, so they wouldn’t always be acting on the same part of the signal.

I’ve never even used a Fairchild on a track that ended up getting released, yet here they are, in a picture, on my website.

Some of the signal maybe be uncompressed, some parts compressed by only one compressor, and some by both. Often the second one is releasing while the other is holding, sometimes one can be attacking while the other is attacking or releasing, so the interplay becomes quite complex.

So, the result of two compressors in series and both set to 4:1 - while theoretically identical to one compressor doing 16:1 - is usually an extremely non-linear form of compression, and one that could never be achieved by a single compressor set to 16:1.

Sonically, serial compression tends to be more transparent because the attack of the second compressor is often masked by the simultaneous movement of the first - and because the compression thresholds aren’t the same, the multiplied-ratio compression is performed on a smaller part of the signal than it would be if a single compressor were doing all the work. Read that again and you’ll see that it does make sense!

Obviously, there are no rules – you can place the slower compressor first, letting transients slip though, and then deal with them with a second, faster compressor! I do this all the time when tracking in my own studio, using a DBX 160XT (in over-easy mode) first as an insert, followed by an 1176. Or use two fast compressors, two slow ones, whatever. Take care not to over compress using this technique, just experiment and have fun.

As a parting bonus: you might find that when using serial compression on a vocal requires more de-essing than you’re used to applying. But that’s okay! It doesn’t necessarily mean you’ve ‘over-compressed’ if that’s what sounds good to you, and you want a really upfront sound. Just use the method described here to go in and turn down the sibilant parts - and you might even find that you need to use a de-esser somewhere towards the end of your plugin chain, as well. Don’t worry, if you think it sounds cool, then it’s all good.

With love,

Ed 🎚🎛❤️