|
Compression Tutorial
|
View this Thread in Original format
| kitphillips |
I wrote this tutorial, because questions are always coming up, and I thought it would be good to consolidate all the knowledge in one place.
VERSION 2
Introduction
Compression and limiting are the most common and most misunderstood tools in the music industry. It is really not a complex concept, but people for some reason have great difficulty understanding how compressors work, and their potential to craft a mix.
I’m writing this tutorial mainly for people who are new to compression, people who are experienced will probably find it oversimplified. Most of this material is my own opinion, and as always there is more than one way to do things. But the techniques I’m suggesting are a good starting point. There is no way I can summarise the whole field of compression in a reasonable amount of space, but I can get through some very simple basics and give you an idea of what you’re doing and where to start.
I’m also dealing mainly with software (although of course some techniques are applicable to hardware). I should emphasise that this is the way I do things, it’s not the only way, not even necessarily the right way. I’m not talking about specific ways to use compression on specific instruments, I’m talking in general terms.
Because we’re talking generally, I should add that every compressor has different conventions. When I talk about increasing an attack control, I mean giving it more time before the attack sets in. Likewise, when I talk about raising the threshold, I mean allowing higher levels through before the compressor engages.
Compression
Compression is one of the most commonly used effects across all genres of music. It has been prevalent since the earliest days of studio recording, and remains one of the most powerful tools in a producers arsenal. It has applications in broadcasts, mixing, mastering and sound design.
Compressors basically control the dynamics of program material, they react in set ways to the dynamics of a song, and adjust the gain automatically. You can usually achieve the same effects as a compressor by automation of volume, but compression makes the task easier because you don’t have to ride faders or draw in automation.
Two of the most common ways to use a compressor, are to add “punch” by strengthening the initial transient of a sound, or alternatively, to bring the overall level of a sound up, removing transients to make the overall level higher. We’ll demonstrate how to achieve both of these effects in a moment, but first it’s important to understand exactly what a compressor is doing.
When a compressor is placed on a channel, it does nothing. The compressor will only kick in and effect the signal when the signal clips the threshold, which is the point at which the compressor begins to act. Once the signal hits the threshold, the compressor will react by making it “harder” for the signal to increase in volume. The amount by which it makes it “harder” is dictated by the ratio control. For example, if the ratio was set to 3:1, then a 3 decibel increase in the input, would be trimmed back to a 1 decibel increase at the output. In the case of a limiter, the ratio is a fixed inf:1 which means that no matter how many extra decibels you pump into the compressor, it won’t make the output any louder at all.
It is very helpful to have a compressor which has a graphical display which shows you where the level is and where the compression is kicking in. Sonalksis and Waves compressors both have this, and it helps greatly in understanding how compression works, because you can see the resistance the compressor is applying to the signal as it tries to get louder.
Controls
So, a quick summary of the controls and some of their common uses (and abuses).
Ratio: Compressors don’t have a simple knob which controls more or less, this is what makes them tricky. The ratio and the threshold both help to control this balance. A higher ratio essentially means more compression, the signal will have to struggle harder to get louder.
Threshold: the point at which compression starts. This should be adjusted down when dealing with a lower level signal. This dictates how loud the compressor lets the signal get before it starts pushing down peaks. If the compressor is triggering too often, give it a higher threshold. If its triggering too little, give it a lower threshold.
Attack: how quickly the gain comes down when compression starts. So, the compressor might not let even the smallest bit of signal through before compression, or it might let the initial transient go for quite a while before it starts compressing. Higher values give more punch, but also result in less headroom, so you can’t turn the track up as loud.
Release: how quickly the compression stops once the signal falls back below the threshold. A value too low will produce “pumping”, a value too high will mean that parts of the material that are soft to start with will become even softer as the gain reduction (which NEEDED to be applied to the loud bits) will also be applied to them. Sometimes pumping can be a good thing.
Input Gain: Essentially the same thing as the threshold, bring the input gain up, you’re basically bringing the threshold down. This was important for gain staging in the days of analogue compressors, it is not important now with software. If your clipping the input of the compressor, it will not help you, since it comes after the input of the compressor by necessity. If you have an analogue modelling compressor, it may be relevant so check your manual for the compressors operating ranges.
Output gain: compression doesn’t make your track louder, it makes the peaks softer. You then need to bring the output back up to compensate. The end result has more even dynamics because the peaks are quieter and the overall volume is raised. Output gain brings the gain back up.
Knee: Only present on some compressors, it’s difficult to explain how it works. It essentially makes the compressor kick in harder or softer. In technical terms it allows the ratio to come on more gradually. For example a value of 10 Db will mean that the ratio increases gradually over a space of 10 Db. This control is your friend if you want the compression more subtle and less in your face.
Common tricks
Now we’ve covered the basics, I’ll show you how to achieve the common effects from a compressor.
Overall volume increase: This is the most obvious way to use a compressor, and will bring a track more to the front of the mix. This is done using a short attack and a release long enough to avoid pumping. A lower ratio is preferable and a threshold which ONLY affects the transients is important. Then bring the output gain up and voila, you have a more even volume across the track. This means that the entire thing can be made louder without clipping, and psychoacoustically, the track will be stronger. This technique is good for sustained sounds, but will not help you if the entire signal is a transient, such as for plucks, drums (splashy cymbals can make use of this technique) and other similar highly transient sounds. Also, if you have too many elements in the mix using this technique, it’ll lead to listener fatigue and a muddy and crowded mix. Its best used sparingly.
PUNCH: This is the more creative application. You can use this technique to bring a track forward in the mix at the points where its transient is strongest. It actually results (counter intuitively) in MORE differentiation in dynamics. What happens, is that you set a hard ratio and knee, then a long attack and a quick release (to create the dreaded pumping). The signal then activates the compressor, but the compressor waits until the transient is through before clamping down. By the time it starts compressing, the transient is through and the rest of the signal cops the volume reduction. This is only really useful in cases where the material has a distinct transient, sustained sounds won’t work. But when people talk about adding punch (especially to a kick) this is what they mean. Because you have more differentiation in dynamics, you’ll have a pumpier mix without necessarily filling it up too much, since certain elements will only cut through and grab attention when they’re needed. Its handy for this application to have a separate limiter put after the compressor (some compressors have them built in) to catch that initial unaffected transient, since it could very well be sending the channel into clipping.
Limiters
Limiters are essentially compressors without an adjustable ratio. They have a constant, fixed ratio of infinity: 1, which can make them sound extremely “hard”. When I’m talking about limiters I’m mainly dealing with multiband limiters which are usually placed last in the effect chain on the master channel. I’m not talking about maximisers which are sometimes placed on individual tracks. I’m writing this with specific reference to the Waves Plugins, but it may apply to other limiters.
Limiters were originally used mostly in radio broadcasting. Today, they are often (over)used to provide greater overall levels in an attempt to make a track sound better. Psychoacoustic research has shown that a track which is “louder” is always perceived as “better” and grabs more attention on radio, thus a volume war exists to see who can bring out the loudest track. A track which is quiet has the ability to kill a dance floor (heaven forbid a DJ use those little pots marked “gain”), or bore a radio audience. For this reason, it is advantageous to master a track for radio differently to the way you might master it for a CD.
Limiters can also make a track flat and lifeless, at their worst they can make a track too pumpy and hurt a listeners ears. It’s important to be aware of this.
Limiters do have their place in the mastering process, but care should be taken not to master every track with the same settings on the limiter, some tracks may benefit from being squashed a great deal, others will suffer. Waves make excellent limiters, and these are pretty much what I’m referring to here, but these techniques should apply to any multiband limiter.
Many compressors have basic limiters before their output, this is to catch any transients which high attack settings may be letting through. This is important and very useful. These limiters are basic and single band. But they are activated for only a split second, so they don’t cause too much listener fatigue and won’t destroy too much dynamic range.
Limiting is basically a master effect, you don’t usually use it on tracks (of course you can....). When setting the limiter, simply fast forward to the loudest parts of your mix, and set the limiters ceiling to your target (usually -.2 db or something). Then begin bringing down the threshold till you see limiting starting to happen. Once this happens, adjust it to where you think it sounds good. Then, look at trying to organise your EQ on the limiter. Multiband limiters (which are far superior for reasons I don’t have time to justify) usually have a priority for each channel. If your noticing you’re losing a lot of a certain frequency, or want to bring a certain frequency out more, then raise the priority of that frequency’s band so that it takes priority over the others. This frequency will now take priority over others in the mix when limiting needs to be done. Most limiters also have an EQ which you can use to modify the final product a little, this acts before the limiter usually, so you may have more gain reduction at those frequencies (nullifying the EQ’s effect) unless you up the priority and the frequency gain at the same time.
Sidechain
Side chain is an easy concept but difficult to implement because of the lack of attention given to it by VST developers. It’s a very common effect, especially in electro and house, where a bassline is often sidechained to the kick.
Sidechaining is basically where the level of one track controls the level of another, so when the kick peaks the bassline is compressed. Obviously, in a normal compressor, the signal is sidechained to itself, so when it peaks it reduces its own levels. In this case, the threshold actually refers to the control signal, the main signal drops.
Most Sidechain compressors also include an EQ to help only trigger the compressor when a certain frequency of the control signal peaks, rather than the whole signal.
Compression Traps
Some quick tips:
• Compression doesn’t usually need to be applied to synths, use the ADSR unless you’re after a special effect. If the dynamics are uneven, play the part better or edit the velocities. Some people might complain that as the filter envelope opens up it causes a volume jump. This is a good thing, but if you really want to completely annihilate all dynamics, you can link the filter ADSR inversely to the volume. Meaning that as the filter opens the volume decreases, keeping the level equal.
• Samples are usually already compressed, and won’t want much more.
• Compression comes before a reverb unless you’re after a special effect. Otherwise you wind up with the reverb being very present in the mix and the reverbs decay being negated by the compressor and possibly drowning out the program material.
• Don’t use a single band compressor on a whole mix, use a multiband, or even better, don’t use one at all. That’s my opinion, some people do it. Many people screw it up.
• It doesn’t matter whether you EQ or compress first, its really personal choice. It does have obvious implications, but there’s no technically correct way.
• If it doesn’t need it, don’t use it, you hear a lot of people saying that they just stick compressors on every channel. Would you do this with any other effect? Compression is best used on channels with high dynamic range like guitars and other recorded sources.
• When you use compression, you aren’t adding something, you’re destroying dynamics, be aware of this. Dynamics are a good thing, they aren’t there to be squashed out of existence for your convenience. |
|
|
| Sonic_c |
Good stuff mate I love it when people who know what they are talking about post easy to follow tutorials.
Thanks |
|
|
| cenik |
| quote: | Originally posted by kitphillips
• It doesn’t matter whether you EQ or compress first, its really personal choice.
|
I just thought I'd post what CM has to say about this issue:
Thanks for the write-up :) |
|
|
| kitphillips |
| Yep, but then theres the argument that to stop the signal from peaking and clipping you should put the EQ before the Comp, so the comp catches the changes you've made and limits them if neccesary. Its a field I don't feel the need to get into, people should do it the way they feel comfortable with I guess. |
|
|
| palm |
| what about just lowering the vol instead? |
|
|
| Sonic_c |
| This processing stuff is more of a science than the actual creative bit!!! |
|
|
| kitphillips |
| I should amend this since I got the ratios around the wrong way. should 3:1 NOT 1:3... I got taken apart for it on another forum. |
|
|
| kitphillips |
| Someone on another forum said that the best way to choose which order to put the EQ and compressor in is to work out whether your doing subtractive or shaping EQ, if shaping (additive) then put it after so that the compressor doesn't false trigger because of your EQ enhancements. If using subtractive EQ, then put it before, since you'll be cutting the level. I thought it was really well put so maybe it'll help you guys ;) Still does come down to personal preference though. |
|
|
| lenieNt Force |
| quote: | Originally posted by kitphillips
Someone on another forum said that the best way to choose which order to put the EQ and compressor in is to work out whether your doing subtractive or shaping EQ, if shaping (additive) then put it after so that the compressor doesn't false trigger because of your EQ enhancements. If using subtractive EQ, then put it before, since you'll be cutting the level. I thought it was really well put so maybe it'll help you guys ;) Still does come down to personal preference though. |
Thats something to think about. Good point..
Thnx for posting that here. |
|
|
| kitphillips |
| New version posted, mostly small qualifications to touchy points to keep really pedantic people happpy. But I haven't updated the actual file. |
|
|
| MrJiveBoJingles |
| Good tutorial. |
|
|
|
|
