[Teetow]

(Complies to all versions)

Well, hello there!

I hope y'all survived the Nyquist Mirroring tutorial without too much brain damage. This time I'm gonna go a bit more easy on you, and I've even prepared some listening examples. Enjoy!

What is a compressor?
The compressor is for amplitude what the autotuner is for pitch: it compensates for imperfections and mistakes caused by humans or technology. With the use of rather simple components (or algorithms, in the software domain) one can construct a very versatile audio processor, probably one of the oldest in modern music production.

How it sounds
The idea of the compressor is rather simple: When the amplitude of the signal goes above a certain level, the compressor automatically lowers the gain to compensate. When the signal weakens, the compressor re-adjusts the gain back to its normal level.

Let's look at a short sample of a piano. There's a link to the corresponding mp3 below.


Piano example A1 (listening example)

Please note that this piano is normalized, meaning that the amplitude has been increased until the sample's highest peak reaches 0db. Still, there's an awful lot of headroom in the sample.

Now, if we were to apply a compressor to this track, it would look like this:


Piano example A2 (listening example)

OK, so now the volume actually got lower, but we're only halfway through. Take a look at what happens if we normalize the track again:


Piano example A3 (listening example)

There's almost no headroom left, and to an untrained ear, the signal still sounds pretty much the same. Please note that I exaggerated the compressor on purpose, just to illustrate its use.

How it works
The basic compressor has four parameters: Threshold, ratio, attack and decay. Most compressors also feature a post-gain knob, to make up for lost amplitude. This image shows the effect of altering each of the four parameters. The dark-blue area shows the original signal, the purple line is the compressor's response.


Threshold
The threshold is the limit for the compressor's "reach". As seen by the picture, increasing the threshold (red line) will make it less sensitive to changes in the material. This is marked with the yellow line. A high threshold is very common when you use a compressor just to make sure that the signal doesn't clip and become saturated, for example when recording (not mixing) drums or rhythm guitar.

Lowering the threshold will make the compressor more reactive, which makes it suitable for subtly narrowing the dynamic range of a piano or a solo guitar.

Ratio
The ratio denotes the "power" of the compressor. In the picture above, the ratio has been decreased from its original setting, and as you can see, the compressor responds more subtly, making the effect less audible.

Ratio is measured in in dB, and refers to the between input and output level. A ratio of 4:1 means that when the input level increases by 4dB, the output level of the signal above threshold will only increase by 1dB.

Attack and decay
So far, we have only discussed the compressor as a tool for smoothing out a signal with a lot of dynamics in it, e.g. to "chop off the peaks". By playing around with the attack and decay settings, we can achive quite the opposite effect and enhance the attack or "snappiness" of a sound.

Take a good look at the "longer attack" setting in the picture above. It might not be obvious to the eye, but the output from that setting would have a very distinct punch. Just think about it: if the compressor waits a few milliseconds before lowering the volume, for a short while the original signal will play through att full velocity, leaving the signal's sharpness untouched. When the compressor finally lowers the volume, this will further give the feeling of a very distinct sound.

The decay parameter (sometimes called release) works in a similar fashion, but affects the time it takes for the compressor to re-set the gain to normal. This, again, is useful to make the attack of the sound stand out more clearly. Bass and kick drums are a good example of where compressors are used to make them more snappy.

Using the compressor
Let's bring up one of my hobby projects, a friend of mine named Alex Benitez, performing his song "without you". This excerpt is from a short bridge just before the final chorus.

Full mix example A1

Oops. Something's not right. Aah, all the compressors were switched off. Here, try this one instead:

Full mix example A2

Much better. Now that you know what compressors can do for your song, let's get into detail. We'll go through the mix, instrument by instrument.

Drums
OK, here's how the drums sound without compressors: Drums example A1

I don't know about you, but that's one of the most boring snare drums I've ever heard. It totally screams "Hey, I'm a sample CD!" at you. I would like to boost the "snappiness" and bring out the tail of the drum, without having it dominate the mix totally. How do I do that?

Well, as you remember, the snap comes with using a non-zero attack time. On snare drums, where the lowest interesting frequency is about 200 Hz, we can have a relatively short attack and still get a good, clear attack. Let's try 20 ms just for a test, but don't be afraid to go as low as 5 or 10 ms, if it sounds good.

Next, the tail. This means that we need to bring the gain back up as soon as possible, and that in turn means having a short decay. I'd go as far as say 0 decay works best for this particular sound, but anything up to 200 ms is a safe bet. We'll settle for 100 ms.

The kick gets a similar treatment, since we want pretty much the same effect (more snap, more "room"). The hihat, due to more hi-frequency content, gets an even shorter attack and zero decay.

The crash cymbal is a bit different, though. It is a real pain in the ass to mix a song where the drummer plays a lot of crash cymbal, since the signal level can go as high as 400% at the beginning of a measure where there's a crash, and you would still like to push it higher up in the mix to avoid that chopped-off sound. The solution? Compress the living crap out of it!

On this particular song, I went for a zero-attack, zero-decay setting with a VERY low threshold (-24 dB to be exact) and a rather subtle ratio, say 4:1. This will take away that very annoying attack, but make sure that the crash ring-out doesn't disappear into the mix halfway into the next measure.

Now, let's take a listen. I recommend that you load up example A1 and switch between the two several times, each time focusing your ears on a different drum. It's hard to hear all the changes at the same time.
Drums example A2

Hey, how about that! A lot more room, more power, more realism. Also, note that I left the tambourine completely untouched, since it is a background instrument, and would only get in the way of the hihat if we were to compress that one too.

Bass
Being a bass player since 10 years back, I always considered my technique to be adequate. But the first time I heard myself play on an unmixed song, I almost fell out of my chair! One second I could hardly hear the bass, the other it dominated the entire mix. At first I though there was something wrong with the instrument itself but the recording engineer explained to me that this was perfectly normal, and that even the most skilled bass players needed a compressor or two to get that professional sound I was used to hearing on the radio.

The trick to compressing a bass is to think about its role in the whole song. If you listen to a soloed bass track from a final mix, you notice that it sounds very processed, almost like a synth. But once you kick in all the other instruments, you hear that a well-compressed bass track is essential to create a "stable" mix. The bass, along with the kick drum, comprises the foundation of the entire song, and as such, cannot vary in volume.

Let's hear the uncompressed bass track from Alex' song: Bass example A1

Now, that didn't sound too bad, did it? Well, scroll back to the Full mix example A1 and listen to the bass. See? We just can't have that much dynamics in the bass, it makes the whole mix unstable. Who you gonna call? Mr. Compressor!

Now, with bass, you need to remember that it's role is to work together with the bass drum. And since we mixed the bass drum with a little "snap", we'll do the same to the bass, otherwise it will sound dull. We'll use more or less the same settings as for the kick, meaning a medium threshold and ratio, a 20-30 ms attack and a short decay. Voilá! Bass example A2!

Again, don't just listen to the individual tracks. Compare them to the fullmix example and hear what effect it has on the final mix.

Guitar
Now, this is not my expert area, and neither is this song a very good example of how guitars should be compressed. There are so many different guitar sounds and playing styles that you simply have to listen and decide from mix to mix what works best.

In this song, I lined the the guitar directly into my soundcard rather than miking up an amplifier, which gave the guitar track a very direct and sterile sound. It also made the guitar very plucky with a huge variation in volume. Guitar example A1

Now, apart from the organ which is put very far back into the mix, the guitar is the only thing that makes the actual accompaniment of this particular song. It must be heard, but can never get in the way of the vocals, or any other element for that matter. The solution was to apply VERY heavy compression to it.

I guess you're starting to get familiar with the settings by now, but here they are anyway: Low threshold, medium ratio, no attack and a short decay. Guitar example A2

You'll notice that all the accentuations of the guitarist are still there, but that it's much easier to manage the guitar track in the full mix (I trust that you've already scrolled back and played the full mix examples A1 and A2 back a couple of times, listening specifically for the difference in the guitar track)

Vocals
OK, this is the final part. First, I'm gonna embarrass Alex a bit by playing back the uncompressed vocal track: Vocal example A1

Listen to the phrase "was my ego". You'll notice that Alex's voice trails off at the end, as he runs out of breath. This is a common problem with singers, and instead of having them retake each phrase with a full lungs, we just compress him to even out the volume. Now, compare that track to this, Vocal example A2, and you'll see what I'm talking about the compressor being a vital tool.

The effect of compressing vocals might be a bit more obvious if we insert the un-compressed vocals into a compressed mix. You'll also notice that we've jumped forward in the song to the last chorus, where all musicians are doing their best to push the song to its climax.

Chorus example A1

Do you hear what he's singing? I sure as hell don't. Let's compress him, and his backup buddy too (which is actually himself). Low threshold, low ratio, quick attack, short decay, mix in a kettle, add some powdered frog legs and... Hocus Pocus! Studio magic!

Chorus example A2

Conclusion
While the listening examples in this tutorial are very exaggerated, and you might not agree about the music style, they serve their purpose to point out one thing: It's not very important how the instrument sounds by itself, it's how they all add together and form a mix. Small variations in volume in all the instruments will contradict each other and turn into a chaos that might be uncontrollable to an engineer, and unpleasant to the listener. The compressor exists to prevent that chaos, and to aid a good producer in getting the sound he wants without doing retakes and heavy editing.

Also, you might think that this is history, and that it only applies to recorded, or "live" music. I'll let you in on a little secret: Apart from vocals and guitar, every single note in this song was generated by a sampler. Drums, bass, organ, even the tambourine are computer-generated. Their levels are kept stable, there are no "human errors", each snare drum beat is given at exactly 127 velocity unless I chose to program them differently. Still, I would never be able to accomplish this sound without the use of compressors.

Epilogue
OK, I know this has been a lot of boring text. I tried as hard as I could to keep the text volume down, and to add as many illustrations as possible, but production theory IS a lot of learning and explaining. If you have gotten this far in less than an hour, I congratulate you. It took me five years to learn it, and I still have a long way to go! =)

I suggest you open up your favourite music software and start experimenting, apply compressors to everything you can think of. You'll probably overdo it the first few times, like I did, but at least you'll then learn when NOT to use compression... Which is even more important.

However, it is beyond the scope of this tutorial, and definitely something you'll have to find out for yourself!

Until next time!

Johan Althoff
Sound Designer
Starbreeze Studios
_________________
----------
Hap hap!

[pahamoka]

This tutorial was pretty usefull for me, although the examples wasn't really the best I could think, especially that piano example which was even worse than the original one. (Ok, I know that the hissing can be edited away with a simple lo-pass filter, but still... )

ps. The most usefull example was that vocal example. Now I know how they can make a singer out of anybody!

[st3vie]

Nice tuto, Teetow!

I "learned a few things"... already knew the global things off it, but still,
very useful, to get more detailed information, to even understand it better.

Thnx again!

-=Stevie

[st3vie]

I hope some ppl that didn't know anything or just a bit about this,
can comment on the tutorials, to get a better insight of the effectiveness.

-=Stevie

[Teetow]

Pahamoka: The piano example was how a compressor sounds, not how it should be used. Still, there might be times when you have to compress a piano track that hard, if it's in the middle of a rock mix for example. Maybe that should be pointed out in the tutorial, though.

And, I never worry about hiss, that's 90's thinking. You can't hear it through all the other instruments anyway. Applying a lopass filter to get rid of it would most likely destroy the piano sound alltogether =)
_________________
----------
Hap hap!

[pahamoka]

Well, let me try to besten it a little...

Well again, I might try it a bit later, as my computer seems to be fucked up at the moment (goldwave didn't like to load up for example)

[DeiFuz]

thx for the tutorial!
very useful and instructive
at beginning I couldn't hear the difference between the mix with & without compression, but with the exmples now I can hear notable differences.
I will use compressors for my own band.
Thx!!

[skagen]

Wow, very good tutorial. Maybe I can fix up my chaos songs when compressor will be implemented in Skale
By the way, is that you singing?

Thanks, and you are god!

[Teetow]

Skagen: nope, that would be my pal Alex Benitez. He wrote the lyrics and the melody, I did the arrangement and sequencing, and arranged part of the backup vocals.

We recorded him along with his guitar, straight into the computer, without metronome since Alex isn't used to that. I then re-cut the tracks so they would fit the BPM we had set (there are some glitches there that you'll notice if you listen carefully) and started adding drums and stuff. A lot of work, but it's good practice.
_________________
----------
Hap hap!

[Mind of Illusion]

Really good tut, I've learn lot of things ... but where is the complete song ? sounds really good to me

[DeiFuz]

Yes, I want to hear the whole song too.

[Teetow]

Sure thing. Here it is:

http://teetow.shells.se/music/alex_benitez-without_you.mp3
_________________
----------
Hap hap!

[Tranceparent]

Hey, that tutorial was really helpfull Teetow, thx.

There's still a lot to learn about this subject for me, but this was a good start.

I experimented a little with compression on some drums in Cool Edit, but does anybody know where to find a nice free compression vst?

[Sportfreund]

hi teetow!

thanks for this great tutorial. i already used compression when mastering in t-racks, but didn't know exactly how it works - i just compressed the whole song

Now i got the idea how it works. Two days ago i mixed a song in cool edit pro 2 and the result is much better when using different compression on the individual instruments. The drums and the e-guitar make pressure and the bass guitar suddenly rocks

So thank you again for this clear description! I hope there will be more tutorials from you!

greetings,

sportfreund

[Templar]

Big thanks to teetow for a very useful and clear tutorial.

It's always nice to learn something new. I actually knew almost nothing about this before reading your tutorial. Now I'm off to experiment