Davide Carbone Demystifies Audio Compression

Share Comments  5s

Compression. In my experience it's probably the most misunderstood dynamic effect in music production. What is it for and how do you use it? I'm going to explain the basics but I'd also like to explain compression in practice rather than just in theory.

 

Firstly we need to understand what it does. The best way I like to explain compression is that 'it attenuates transient peaks', that is, it smooths out peaks in your audio signal you can't hear. Nasty stuff that takes up valuable headroom. I use compression throughout the track on individual channels sensibly in order to get rid of unnecessary transients whilst retaining dynamics. That then gives me a smooth, dynamic mix which I can feed in to my final mastering chain to achieve loudness. For me, music production is all about the balance between dynamics and loudness.

 

The three most important things to look for on any compressor are the Threshold, Ratio and Gain Reduction.

 

The threshold sets the level at which the compressor kicks-in and starts changing the dynamics of the recording. So for example, if you set your threshold at -20 dB, everything below this level will not be affected by the compressor. But everything louder than this level (-20 dB) will be compressed.

 

How much will the signal be compressed once it's gone over this threshold level? This is controlled with the ratio. The higher the ratio, the more compression there is. The graph below demonstrated various degrees of ratio. If the ratio is 1:1, there is no compression at all.

 

If the ratio is set at 2:1, for every 2 dB of sound that goes over the threshold, you get 1dB of output above the threshold. So if the signal goes over the threshold by 10 dB, the compressor reduces this signal so it's now 5 dB over the threshold.
If the ratio goes up to 8:1, for every 8 dB of sound over the threshold you would get 1dB of output above the threshold. So if the signal goes over the threshold by 16 dB, the compressor reduces this so only 2 dB goes over the threshold.

 

Now the Gain Reduction will tell us exactly how much of the signal has been attenuated (reduced) in decibels.

 

This graph details how Threshold and Ratio work, note the difference between compression and limiting. Limiting has very little respect for dynamics and will just eliminate the peak entirely:

 

 

 

So how do we use it in a mix?

 

As a guideline set the ratio to 2:1, now adjust the threshold until you hear a pumping effect. Now back off to the point that you don't hear any audible compression, minus 10 - 20 are ideal threshold settings. Remember to not to go crazy with compression as you may suck the dynamics out of your track. Try not to remove more than 6dB on individual parts. Light compression over each channel with a good compressor over the entire mix would be ideal.

 

When recording vocals or instruments I always like to do two things: low cut and compress. Obviously I wouldn't low cut a bass guitar or a kick drum but everything else I'll generally remove anything under 75hz on the way in. I'll also add some subtle compression, nothing audible just something that may reduce the signal by a few dB. This way I've collected some of my character from my EQ and compressor and given myself every chance to have a good signal to further process - or not!

 

Here's an image of an example mixdown. Notice most channels feature compression at the end of the chain. That's the way I like it because any EQ or effects I have added may have reintroduced some transient peaks. Remember, the key here is to be subtle - don't kill the dynamics. Finally I've added some multi band compression and limiting only on the output. The purpose of this is to make the track loud! Loudness and dynamics, I believe you can comfortably have both.

 

 

So what type of compressors should we look out for? Well it's a matter of software or hardware. I like to classify my compressors as being either precise or warm. One is like an exact surgical instrument, the other adds character and colours the sound. Neither is better it's just a matter of preference. In terms of software compression they are all very much 'precise'. The only compressor that really does add warmth and colour (in a good way) is the Waves Renaissance Compressor. In terms of hardware there are literally hundreds of options. The best way to test them is to take a piece of music to a store with a listening booth and choose the compressor that makes your track sound best!

 

Hope all this helps. I'm happy to answer any questions you may have, just leave a comment below. Of course, you are welcome to join our very popular Production Masterclass at the School of Synthesis in Melbourne where we do exactly this and more for six weeks!

Comments

Flexotron says:

That was real helpful. Any suggestion for a master compressor for a thousand bucks? Thanks.

Davide Carbone says:

As you intend it as a master compressor I would look for a soft knee compressor that has some warmth and character. Look for an original Joe Meek SC2 or SC3. Avoid any of the recent Joe Meek series, they are a darker green and not made by Ted Fletcher. Be patient and you'll find an SC2 as used by Prodigy through to Coldplay for less than 1K.

shaun.keyt says:

Hi Davide,
Do you ever insert an EQ after a compressor, eg to further manipulate the compressed signal.

Davide Carbone says:

Yes I do Shaun, particularly when creating an effect. If I re-introduce peaks I'll happily add another compressor. I do this with side chaining often.

Peter says:

All pretty clear although very difficult to 'hear' the differences between digital emulations of compressor types (FET,VCA, Opto etc)