Demonstrating the Waves J37 analog tape emulation plug-in and comparison with a real tape recorder
"There is background noise in my studio. Should I use a noise-reduction plug-in?"
Audio problems at the BBC - TV drama audiences can't understand what the actors are saying
Make an attention-getting lo-fi introduction for a track
A simple 8-mic drum mix, with video
New vs. old guitar strings: Part 2 - The case for used guitar strings
A brief introduction to acoustic treatment
This voice over studio looks like something out of Monty Python
Setting microphone preamplifier gain to achieve both adequate headroom and a good signal-to-noise ratio
Can you hear the subtle effect of the knee control of the compressor? (With audio and video demonstrations)
Subscribe to access our latest, up-to-the-minute articles with hints, tips and adventures in audio in the weekly Audio Masterclass Newsletter.
You'll need to know the basics of compression already for this to make sense. Got that? Then read on...
The theory is simple. If you set a hard knee then the transition between no compression and full compression is sudden. As the signal level rises, there will come a point where it is compressed at the full ratio that is selected. This will be exactly and instantaneously at the threshold level you have set.
Set a soft knee however and the signal will start to be compressed at a low ratio as it approaches the threshold level, then at a higher and higher ratio as it crosses the threshold, eventually reaching the full ratio that you have set.
The purpose of the knee control therefore is to allow the compression effect to be more obvious (with a hard knee) and less obvious (with a soft knee). Which you choose depends on what you want to achieve - obvious compression or discreet compression.
The problem is however that the knee control is very subtle in effect, so you have to listen hard to hear what it is doing. Or to put it another way, you have to increase your aural awareness until it becomes easy for you to hear.
So here is a demonstration of the difference between hard knee and soft knee. And you will have to listen hard, but once you 'get it', then the effect of the knee control will become very much clearer.
The example here is a short clip of It Would Be So Easy by Jess Penner, sung in this case by Charlotte Roel. It's a very short clip so that you can listen repeatedly and catch the small differences. It also has a sudden transition between quiet and loud, which makes the effect of the knee control more obvious.
The compressor is the standard plug-in that comes with Pro Tools, but any compressor that has a knee control would be suitable. The settings chosen help bring out the differences that can be achieved by varying the knee. You can ignore the side-chain controls in the upper right as they are not active.
Example 1 is the original signal with no compression (the plug-in is bypassed), recorded using a vintage Neumann U47 vacuum tube microphone. It is set so that the peak level is -3.7 dBFS. That precise figure isn't important, but it will be used all the way through the examples for consistency of comparison. This has been done using make-up gain in the compressor plug-in.
Now that you know what it sounds like without compression, here is the version with the knee as hard as it can get...
What you should look for here is the shape of the compression curve. A straight line at a 45 degree upward angle means that there is no compression. The more the angle of the curve tilts downwards, the greater the compression ratio.
Here we see that the curve is comprised of two straight lines - one indicating no compression, the other almost horizontal and indicating a very high compression ratio. The transition is sharply angled, which is visually metaphorized as a hard knee.
So let's hear what the soft knee sounds like...
This is where you might easily say, "It doesn't sound much different." You're right, the difference is small compared to the degree of difference you can make with any of the other controls, but it is important. Here's what to listen for...
In the quieter section before the loud note, the signal is hardly compressed at all, and then only momentarily. Compared to the uncompressed signal there is virtually no difference.
In the louder section, the signal is very clearly compressed. It is easy to hear the difference between this and the original.
So it's either nothing or something. No compression or lots of compression.
In the quieter section before the loud note, it is easy to hear that the signal is compressed compared to the original. In the louder section, the signal is also compressed, but not as much as in the hard-knee example except for the very loudest note. The transition between not-compressed and compressed is smoothed out.
Since this is an exercise in making the difference between hard knee and soft knee as clear as possible, I'll do something that you wouldn't do in any real life application - I'll combine the two examples into stereo.
In this demonstration the hard-knee example is panned left, the soft-knee example is panned right. The peak level is maintained at -3.7 dBFS. Take a listen (preferably on headphones)...
So let's hear what it sounds like...
That will sound weird. But what you are listening for are the differences between the two channels. You'll hear that as shifts in the stereo image. Where the image is centre, the channels are the same. The further the image shifts, the more difference there is.
So what you will hear in the quieter section of the example is that the louder parts are more-or-less identical, but the quieter parts - very short syllables and breaths - shift to the right. This is because the soft-knee version on the right is more compressed and the make-up gain brings the quieter parts up in level.
In the louder section, once again the very loudest parts are more-or-less identical, but this time the greater amount of compression in the hard-knee version brings the less loud syllables and breaths up in level.
Still can't hear it?
OK, this time we'll use a visual aid. This will be the Flux Stereo Tool, which shows the stereo signal as a Lissajous display. Only the centre circle is relevant here. To explain briefly, the louder the left channel is at any moment, the more the display will tilt to the left. The louder the right channel is, the more it tilts to the right. Using this, we can gain a visual as well as an aural appreciation of the difference between hard knee and soft knee.
You may need to watch and listen several times to appreciate the subtleties, which are nearly all in the quieter and shorter snatches of syllables and breaths. But once you start to get it, you can return to the individual examples and hear the audio as it would normally be heard.
In short, if you have a compressor that has a knee control, you need to decide whether the knee should be hard or short, or some setting in between. If you want to hear the effect of compression, then - in combination with appropriate settings of the other controls - the knee should be hard. If you want the compression to be effective but unobtrusive, the knee should be soft.