Audio demonstrations of distortion produced by compressor plug-ins
Today I'm going to take a look at distortion produced by compression plug-ins. Distortion, as you know, is sometimes a good thing, sometimes bad. Mild distortion can be interpreted by the ear as 'warmth'. Taken too far though, distortion can be an ugly sound and definitely unwanted. I had the 'pleasure' of listening to some audio recently that was so distorted I couldn't tell whether it was speech or music. Definitely not wanted.
Monitor health warning (ears too): Tones follow!
Do compressors always distort?
This is an interesting question. Suppose you performed compression on a signal, to reduce its dynamic range, purely by fader riding. That wouldn't cause any distortion surely?
Well no, not when the fader is at rest. But think of a sine wave as it is faded down. Consider just one cycle. As the fader moves down, the first half-cycle is louder than the second. You have just changed the shape of a sine wave. That's distortion.
The thing is though that you would have to move the fader extremely quickly to create any audible distortion. I doubt that it is possible.
But when you bring an actual compressor into consideration, an electronic, digital or software compressor can easily change the gain quickly enough to create significant distortion, more so on low frequencies. For instance, if the attack of a compressor is set to less than 10 milliseconds, it can react to a single cycle of a 100 Hz sine wave, changing its shape significantly and creating distortion that is most definitely audible - if you listen carefully enough.
Here's a clean 100 Hz sine wave...
And here is the same sine wave compressed. Ignore the bump at the start where the compressor kicks in and give your attention to the steady-state signal.
Can't hear much of a difference, other than the bump at the start? Well it may be that your browser is playing the .ogg version rather than the .wav, but there are downloadable .wav files at the end of this tutorial for you to audition in your DAW. Or better still you can try these tests yourself.
You will however hear a difference if I subtract the clean 100 Hz sine wave from the compressed version. Play the file below and turn up your monitor once it is playing. Don't forget to turn down your monitor afterwards. Ignore the remnant of the 100 Hz left behind and try to hear the second harmonic at 200 Hz.
The distortion components are now clearly audible. If you consider that a sine wave is the simplest audio waveform, then clearly distortion will have much more of an effect on a complex signal like music. And where there is simple harmonic distortion, which in small amounts doesn't sound too bad, there is always intermodulation distortion, which does.
In normal use
I just needed to clear up the point about a short attack time causing distortion. From now on where possible I'll set an attack time that is unlikely to cause any problem. If you hear distortion, or see it in the screen shots, then it is the intrinsic distortion of the compressor that no amount of playing with the controls will remove.
The Waves PuigChild is a recreation of the famous Fairchild 660/670 vacuum-tube compressor. In truth, the whole point of using this compressor is to create warmth, so any mild distortion is not a problem - it's a bonus!
To visualize the distortion products of the compressors I'll be looking at I'm using the Blue Cat FreqAnalyst Pro spectrograph, which is a handy tool for looking at frequency balances. Here's what it makes of a clean sine wave at 100 Hz...
Using the same 100 Hz sine wave and typical compressor settings we get this...
I think that the Fairchild definitely lives up to its warmth-generating reputation. It's worth noting that even though the strongest distortion product is more than 40 dB lower than the 100 Hz fundamental, the difference in sound texture is clearly audible. It is also worth noting that according to Waves's manual for this plug-in, the longest attack time the PuigChild is capable of is 400 microseconds, which is available on Time Constant positions 3, 4 and 5, and this short attack could be creating some of the distortion.
A standard DAW plug-in
When a DAW comes with a compression plug-in as standard, then unless it positively claims to add warmth then one might assume that it is a simple clean compressor. Well let's see about that. Here I have used the Dyn3 Compressor/Limiter that comes with Avid Pro Tools.
Like the Fairchild I have used settings that give about 10 dB of gain reduction, which is about the most you would use in typical compression. I've set fairly long attack and release times so that this shouldn't be the cause of any distortion. But you can hear and see for yourself. Once again you should ignore the bump at the start because it is the steady-state part of the signal I'm interested in...
I have to say that when I first noticed this, it came as a bit of a surprise as I expected the Dyn3 to be clean. Having said that, the distortion is well down in level. The strongest component at the second harmonic is around 60 dB lower than the original 100 Hz, so it would be unlikely to be audible in normal use. Even if it is audible, the compression you are using is changing the signal massively more so a little bit of funny business 60 dB down really doesn't matter that much.
A clean compressor?
I could give you more examples, but it wouldn't really make my point any better. My point is, by the way, that distortion often comes hand-in-hand with compression and you should expect it. Sometimes you will want it; at other times you would prefer the compressed signal to be completely clean. But is that possible? Well how about this..?
The distortion here is well down in level. The strongest harmonic is around 90 dB down on the original 100 Hz and no-one is going to hear that! And significantly I have set a short attack time of 10 milliseconds. This genuinely can be called a clean compressor.
It is the Channel Strip plug-in that comes with Pro Tools and, of course, only works with Pro Tools. But it's the cleanest compressor I have available at the moment so I chose it for this demonstration. Doubtless there are other clean compressors around too, but you'll have to do your own tests.
Where there is compression it is likely that there is distortion too. If it's warmth that you want, then that's great. If you just want clean compression then it's likely that a simple compression plug-in won't create distortion at a level that is going to be troublesome. But it is always worth considering that a really clean compressor is nice to have in one's audio toolkit.