One of the most popular posts on this site is “Using compression to add punch, warmth and power to your mix” – a tutorial to help people get started with the basics of using compression.
In it, I suggest a rule of thumb:
If the gain reduction meter doesn’t return to zero several times a bar, you’re almost certainly using too much compression
And I stand by that suggestion – most of the time.
Now, just yesterday my good friend Joe Gilder over at Home Studio Corner put up a post on this subject, and quoted my rule of thumb. And immediately, the Devil’s Advocate in me wanted to post a reply. It got so long, I changed it into this post – I think I see a trend developing !
So, this post is about the times when you might want to ignore that rule – after all, rules are made to be broken, especially rules-of-thumb : )
The exception to the rule
When might you want to have compression constantly happening ?
When it sounds great.
OK, I know – when is that ? The answer often involves a parameter in compression that people don’t talk about much – the knee.
What is the ‘knee’ of a compressor ?
The knee is the area where the compressor changes from ‘not compressing’ to ‘compressing’ – the threshold.
If you use a compressor plugin and it shows a graph, the knee is where the graph changes from being a straight 45-degree line into a flatter line at a steeper angle – where the ratio changes from 1:1 to 2:1, 4:1 or whatever.
Some compressors allow you to adjust the knee between ‘hard’ and ‘soft’. ‘Hard knee’ compressors just have a corner at the point where the curve changes.
This means the compressor switches straight from ‘no compression’ (if the signal is below the threshold) to ‘some compression’ (above the threshold).
‘Soft knee’ compressors don’t show a hard corner at this point in the graph – they show a smooth curve. This means that the change from ‘not compressing’ to ‘compressing’ when the signal hits the threshold is less abrupt.
It can sound smoother and more natural than a hard knee, especially with higher ratios. However it also means that some compression actually happens below the threshold – so for the same settings, you hear more compression happening sooner.
Now if you set your compressor up using my rule-of-thumb, always allowing the compressor to ‘relax’ back to zero dB gain reduction several times a bar, you are basically choosing that the input signal is constantly crossing through the knee of the compressor.
The signal starts out below the threshold, then as it gets bigger it crosses the threshold and ‘hits the knee’. While it’s loud enough the signal is beyond the knee and is being compressed, then it drops back again… and so on.
But some engineers will choose to push the signal harder than this, so that it’s almost always on or past the knee. Why do they do this ?
Some compressors have nice knees
Plain “vanilla” digital compressors do a great job of mathematically controlling the audio signal. That’s the great thing about computers – you can get them to do almost anything you want them to.
But back in the ‘good old days’, when great analogue compressors were built by hand and cost more then a small car, it didn’t work like this. Every component in a compressor (in fact, every item of audio gear) was designed as a loving compromise between quality, performance, reliability and cost.
The essential aim of the design was the same mathematical concept, but the method of achieving it was very different, and often involved lots of trial and error.
And valves, often.
And people, making choices based on taste, not tech.
The end result of this was sometimes compressors which had curves which looked nothing like the theoretical ideal. Maybe the knee was very soft, or started soft and got gradually harder – maybe beyond a certain point the ratio actually started to ease off again, so if you pushed it really hard you would start to hear less compression – the Urei 1176 is a classic example of this in ‘British’ mode, where all the ratio select buttons were pressed in at once !
And this meant those compressors sounded different – superb, in many cases. and some of that unique sound was because they had great-shaped knees.
Some engineers liked the sound of those knee-shapes so much they choose to push the compressors very hard, so that the sound is almost constantly on or above the knee – the gain reduction is almost never zero.
The point isn’t necessarily to over-compress in the sense of “squashing all the life out”, it’s to exploit the sound of the compressor when the signal is always on the knee of the compressor. It’s not a ‘mathematical’ signal-control compression setting as suggested in my ‘getting started’ blog post, it’s the sound of the shape of the knee – but as you’ll often hear people say, if it sounds right…
Does the knee account for all the great sound of those old hardware units ? No. But on some of them, it’s a big part. Even when you’re not compressing really hard, the knee helps determine how the compressor sounds.
What if I don’t have a classic analogue compressor ?
Luckily, the makers of audio software know all about this stuff, and have designed far more sophisticated compressor plugins. So as well as the “vanilla” plugins included with most DAWs, companies like Universal Audio and Waves spend a great deal of time emulating the sound of these ‘nice knees’ in software – and all the valve distortion and other analogue goodness, too – for a price.
In fact, you don’t even have to pay the earth – even the compressors included with Logic include emulations of these more interesting knee shapes – and make sure check out this comparison video of the extremely affordable Stillwell Rocket plugin head-to-head with the UAD 1176 LN here to see what you think.
Do the plugins sound as good as the hardware originals ? That’s a whole other issue ! But if you want to experiment with bringing your compressor to it’s knee – go for it.
Now, in my opinion this is a more advanced compression technique, and if you’re just getting started it might be better to use compressors purely for control until you get the hang of it. And, in my experience this strategy works best on individual instruments rather than a whole mix – but that’s for your ears to decide !
(Just remember – compressors are for managing dynamic range, not destroying it…)
What do you think – do you sometimes like the sound of over-compression ?