There are a couple of different kinds people in the world that operate in a condition of extremes. The first type is the kind of person who turns the stereo up really loud, drives really aggressively and has really definite opinions about EVERYTHING. I’m one of those people.
The other kind of person is the exact opposite. They do everything safely. They live out their lives with precise control: They drive the speed limit, they never eat unhealthy things, they listen to music in their cars at reasonable volumes even when the police cars are not sitting next to them at red lights and they go to “tasteful” movies that win Academy Awards. These kinds of people bore me to tears and drive me nuts.
I mention this because you will run into a large number of recording engineers that fit into the second category.
When I’m mixing or trying out new effects, I’m the type of person who will drive the level of an effect or a track up all the way just to see how nasty I can make it sound and then I back it off so that it fits with the context of what I’m doing. This way you can see just how ugly a plug-in or a piece or outboard gear can make something sound before you use it “tastefully.”
I’ve always felt that you have to revel in bad taste before you can really appreciate good taste. Making cool, new innovative music is rarely about good taste.
So what does this have to do with anything?
Well, today we’re going to talk a bit about Compression. Compression is a wonderful effect that, along with equalization, can totally take your recordings to the next level - but if you use too much of it, it will kill all of the dynamic range of your music and make it fall flat.
In order to understand how compression affects a signal, you have to try using it in an extreme way so that you know what BAD compression sounds like. Then you back it off and use it subtly to make the individual tracks or the overall mix really dramatic.
Both compression and E.Q. can be used to manipulate the realism of an instrument. You will hear me expand on this topic in the future, but I think that multi track recordings should not just be realistic reflections of the combined instruments that are used to produce them. Recordings should sound bigger than life. They should be the audio equivalent of a Technicolor movie where the colors on the screen are much more vivid that they are in real life.
When we are talking about compression in the context of recording, there are two types: Dynamic Range Compression and Data Compression.
Data Compression takes an audio file and discards parts of it that scientists have deemed inaudible due to a phenomenon called Frequency Masking. Frequency Masking assumes that certain frequencies are inaudible because they are “masked” by other louder frequencies. This type of compression is also called “Lossy” compression. Regardless of what anyone tells you, lossy compression formats like MP3’s and M4A’s do not sound as good as music files that have not been compressed. This is mainly because there are lots of intangible aspects to human hearing that add realism to a sound that can’t be measured in hearing tests or in frequency profiles on an oscilloscope…but I digress…!
The type of compression we are talking about is Dynamic Range Compression. Dynamic Range Compression makes soft parts of music louder and loud parts of music softer – and therefore easier to work with and mix together. It makes music more listenable on low-grade consumer music systems like cheap radios, boom boxes and even the limited range speakers on low price point televisions. Additionally, depending on the controls that are available on the particular device or plug-in that you are using, compression can be used to sculpt the shape of the sound and it can be applied to individual tracks or entire stereo or surround mixes.
Compression is getting a bad rap.
There is a huge trend right now to totally squish the dynamic range of CD’s and make them as loud as possible. In doing this, much of the subtlety of the individual instruments and the interplay between them is lost - and mixes can sometimes blur into an incohesive mess. Because of this trend, you will see tons of articles that expound on the EVILS of compression.
The truth is that compression is a really useful tool and it’s only bad if you’re a big idiot and either use it improperly, or use it too much.
Compression is really hard to write about and explain to people.
On of the reasons that compression is so hard to discuss and so confusing is that it is TOTALLY IMPOSSIBLE to come up with any hard and fast rules on how to use it. This is because of all of the variables that exist with the instruments that you are recording, the microphones and the preamps that you are using and whether you are using tape to record the information or whether you are using a computer. Additionally the compressor itself - whether it’s a stand-alone piece of hardware or a software plug-in - will make an enormous difference in the outcome of the sound. As a result, trying to describe when to use compression, how much of it to use, or how to set the controls is a highly inexact proposition.
Compressors have multiple parameters that can be adjusted. Some devices have all of the adjustments listed below; some of them have just a few. In describing these parameters I’m purposely being non clinical – so my descriptions might not be as scientifically accurate as…say an electrical engineer’s description might be, but hopefully your eyes will not glaze over as you are reading the descriptions and the ideas will actually sink in!
- Threshold – Compression can either be very subtle or very severe. The threshold kind of “draws a line in the sand” and controls how loud the signal has to be before the effect kicks in and begins to reduce the volume of the signal. If a signal’s volume crosses that line, it gets squashed down. A low threshold means that the signal does not have to get very loud before the compressor kicks in. A high threshold means that the signal has to have quite a bit of gain behind it before the effect is enabled. As a result, a low threshold means the compression is fairly severe. A high threshold means it is subtler.
- Makeup Gain – Once the signal has been compressed, the resulting signal is lower than the original signal that went into the compressor. The makeup gain allows you to adjust the outgoing signal so that it’s loud enough to be usable.
- Ratio – Controls how strongly the effect acts upon the signal. A ratio of 8:1 means that for every eight decibels of gain fed in to the compressor, the output would only increase by one decibel. A ratio of 20:1 is much more severe. For every 20dB fed into the compressor the output is only increased by one dB.
The final two controls actually shape the waveform
- Attack – Controls how fast the compressor reacts once the threshold has been reached. A slow attack allows quick transient spikes in a signal to pass through the circuit before the compressor has a chance to affect them. This is a GOOD thing for drums because you want the drums to retain the power of the initial Thwack on the drumhead or the cymbal. If the attack was quicker the drums might lose some of their power. There are other instances where you want the compressor to react really quickly to any sudden transient signals and we’ll talk about that when we discuss Limiting a bit later.
- Release – Is the amount of time it takes for the signal to recover from the effect. A slow release might allow a drum to sound bigger, because the compressor draws out the decay envelope and elongates it after the drum has been hit. A quick decay might make the presence of the effect less obvious.
Anytime the ratio of compression goes beyond 10:1 it is usually thought of as Limiting. Limiting is really just a severe form of compression that keeps the level from exceeding a certain point. It’s like putting a trampoline in a room with a ceiling. You can bounce on it but you can only jump so high before the ceiling stops you.
I tend to think of compression as an effect that works in varying degrees of subtlety. Some forms of compression are barely noticeable and some are extremely obvious. In my mind, even when compression is “transparent,” it is still intended to obviously modify the sound of the signal even if it only does it a little bit. The attack and release parameters in particular allow you to shape your volume envelope in all sorts of interesting ways to make the sound BIGGER.
Limiting on the other hand has a vastly different purpose. Limiting is in place to keep a signal from going beyond a certain level and causing distortion in an audio chain while allowing it to retain a high degree of power. Limiting makes our perception of a signal LOUDER.
So…I don’t mind hearing the results and the artifacts that sometimes accompany compression, but limiting should act upon a signal pretty invisibly.
How do you avoid overusing compression and making other mistakes when you use it?
- Well - the ultimate test for any effect is to bypass it once you have set up a sound that you think is good and then say to yourself “Is this better sounding than it was before I put the effect in the chain?”
- Remember what I said about using recording as a means to make instruments sound “larger than life.” If you compress a track and it sounds smaller, you’re probably not doing something right.
- Although it’s a good idea to solo the track that you are inserting the effect into so that you can see what the basic sound is like - the ultimate test is to see how the track sounds in the context of the mix. So – Solo to get an idea of the sound and then look at the overall mix for the big picture of how the effect on the track works in the context of the mix. Tracks that work great in a mix often sound kind of weird by themselves.
- One of the main mistakes that people make with compression is that they use it all at once rather than staging it throughout the process of recording and mixing. The process of using incremental compression can be thought of as a type of GAIN STAGING. So you use a bit when you record, a bit more on the individual track, perhaps a bit more on a submix on a group of instruments like drums and perhaps a bit more over the final mix if it is REALLY needed.
- I generally always use a hardware compressor and preamp combination when I am RECORDING a track. So, before the sound even hits the converter, A LITTLE compression has been applied. By a little, I mean, a ratio lf about 2:1 or at the most 3:1 with a pretty high threshold, a slow attack and a medium release. With this type of compression there is NO UNDO - So it has to be subtle. You should not really be able to tell it’s there unless you really listen.
- If you are working with instruments that have transient level spikes – like drums, DON’T TRUST YOUR METERS to let you know when the signal is overloading. Meters don’t work fast enough to tell you when there is a spike that can cause distortion and ruin a good track. It’s better to record at a lower level…say about 70 per cent of maximum for the peaks and then stage the gain up later in the mixing process.
Here are some Basic Parameter Settings for Compressors. But remember, in the reality of recording there are NO RULES and everything depends on the type of instrument that is being recorded, the kind of music that is being recorded, the signal chain and the overall sound that you want to achieve.
Threshold Ratio Attack Release
Drums Low 3:1 or 4:1 5-10 ms or Faster Medium - 250 ms
Bass Low 4:1 20-60 ms or Faster Medium to Long
Electric Low 5:1 to 15:1 10-50 ms or Faster Long - .5 - 1
Guitar
Acoustic Medium 2:1 - 3:1 10 – 20 ms Medium 250 ms
Guitar to High
Vocals Start High 3:1-4:1 Fast to Medium Depends on Tempo
and work back
Keys Start High and 5:1 5-10 ms or Faster Medium
Work back
- If you are using Compression and Equalization together, insert the compressor in to the signal chain first followed by the equalizer.
- Bass heavy signals will fool the compressor into over compressing the signal because bass frequencies are more powerful than higher frequencies. If you find that no matter what you do, the upper frequencies sound wimpy, you might try this:
- Copy the track that you want to compress onto a new track and then mix the two tracks side by side. The original track should have no compression and the copied track should be squished like a bug, ie – Low threshold, high ratio, long attack, medium decay and makeup gain to bring the signal back to the original volume.
- Mix the two tracks together. The combination of the two might solve your problem. If it does not, you can copy the track a third time, only this time, apply two different kinds of E.Q. to the copied tracks AFTER the compressor. On the first copied track insert a high pass filter and take out most of the bass spectrum and compress the remaining signal to taste. On the second copied track insert a Low Pass filter and remove most of the high-end spectrum and compress the resulting signal to taste. Then, combine all three tracks to get your sound. This entire process is called Parallel Compression and it works great for drums. If inserting the E.Q. AFTER the compressor does not have the effect you want, try inserting it BEFORE the compressor and doing the same thing.
- If you have an instrument that needs to be limited to make it louder and you want to also make it bigger sounding, it’s better to apply two different plug-ins in the signal chain. First insert the limiter with a high threshold and a ratio of about 20:1 or more. As you lower the threshold and increase the makeup gain, the track will become more powerful sounding. The attack and release will be very dependent on your source material, but normally you would use a pretty fast attack and a medium release
- After you have Limited the signal then you can compress the sound to modify it’s attack and decay envelope and make it BIGGER sounding.
So…that’s all for this installment on compression. Send me some questions if you have any. Compression and E.Q. will be a black art that you will be striving to improve for the rest of your life. Get good at them both.
