Music Production 101: Compression

While being one of the most fundamental tools to getting your productions tight, compression is also the least intuitive at first glance. Some compressor plug-ins are fairly basic and non-threatening while others have more knobs and dials than a NASA shuttle launch control.

To add to that, it’s difficult to actually hear what a compressor does if you don’t know what you’re listening for (it’s actually more about feeling, but more on that soon).

I used to be guilty of blindly tweaking until something sounded OK, and even got away with loading some all-purpose preset (e.g. “Aggressive Male Rap Vocal”). However, understanding what any device/plug-in does and what problem it was designed to solve are the keys to breaking this production thing wide open.

Once you break down all of the aspects of the compressor, you quickly see the usefulness and why engineers came up with it back in early days of broadcasting.

The Magic Volume Puller-Downer Robot

A key aspect of mixing is balancing the overall levels of each track that make up the song (vocal, snare, kick, cymbals, rhythm guitars, lead guitar, bass, synths, etc.). Too much bass guitar and the production becomes a rumbling mess. Too much hi-gain rhythm guitar and the drums and low end are washed out.

It would be very simple to just raise the volume of the snare a bit, bring down the kick a notch, up the vocals, balance the loudness of the bass with the kick, and in fact sometimes this really is all you need to do to “mix” if the recording technique is good.

However, things happen during a performance; the vocalist might sway back and forth from the mic during a single phrase, which causes the first half to sound a little soft, while the last couple of words in the phrase get very loud as she leans in to it.

Or, the drummer might play some light ghost notes on the snare just before a heavy-handed fill.

These differences between soft and loud are referred to as the dynamics of the performance or track. The compressor is a dynamics device (thus you find it categorized under ‘Dynamics’ in most DAWS!).

When adjusting the levels of individual instruments in our song, the dynamics—with soft and loud or really loud moments—may create difficulty in raising the volume of that instrument up in the mix.

If we simply raise our vocal track up so that can we can hear our singer’s soft phrases, we will get blasted in the ears when she lets it rip on the loud parts!

This is where the compressor comes in—like having a magical little robot that watches your audio signal and pulls the volume down if it gets too loud at any given time.

A Quick and Not-too Confusing Note About Meters

metering-logic
Figure 1. Logic’s negative-value dB scale, like most DAW’s, is a little confusing to beginners.

A sound/signal in your DAW (Logic, Pro Tools, etc.) is measured on a scale where the dB (decibel) values listed are mostly negative.

By default, the faders in a DAW are usually set to 0 dB. You can raise a track’s fader higher than 0 dB  (+0.1 to +5 dB or so), but this is introducing artificial gain to boost the signal.

You would think that a 0 dB sound would be completely silent. This is not the case when you look at your track’s faders. 0 dB is a neutral position letting you know that no gain is being taken away (volume reduced) and that no gain is being added (volume boosted).

In Logic Pro 9, my go-to DAW, the dB scale on each fader slides from -∞ dB to +6 dB. The tick marks on the level meters in plug-ins (compressors, EQs, limiters, etc.) and on the mixer windows, like the one in Logic shown in Figure 1, will usually read – 60 dB and up.

A good recorded signal (your voice, your guitar amp) will probably be in the -12 dB (on the more conservative end) up to -3 dB (approaching but not yet clipping on the way into your computer).

Check your mixer window and the meter for that track (e.g. your kick drum track), and watch the meter jump up and down to get an idea of what dB level your signal is pegging at.

These values are important for knowing what you are looking at when you have a compressor in front of you. However, this inverted, negative dB scale is pretty frickin’ confusing at first, but the more you work in your environment and pay attention to metering it will become second nature.

Anyway …

Let’s take a look at the stock compressor that comes with Logic. It’s got all of the basic anatomy of pretty much any compressor you will come across in whatever platform you’re running (Pro Tools, Cubase, Reaper, etc.).

logic compressor.png
Figure 2. Logic’s native Compressor plug-in.

There are some parameters here that I will get into on a later post (Circuit Type, Side Chain, Limiter).

Side Note – In the channel strip or plug-in chain, some engineers put the compressor before the EQ, others do the opposite (like me). By my logic (pun intended), EQ’ing the signal changes its dynamics by cutting or boosting frequencies, and I feel like it generally makes more sense to compress after you have determined the EQ settings. There’s really no hard and fast rule, so try it both ways and hear how it sounds!

Now let’s break down the anatomy of the compressor:

Threshold

The Compressor Threshold setting is the level in decibels (dB) that the incoming sound has to exceed for the compressor to become activated and pull the volume down.

The Threshold is the hard line – if a sound gets louder than the Threshold you set, the magic compressor robot springs into action to bring the volume down.

Let’s say your compressor is armed on a vocal track. You’ve set your Threshold to -5 dB. Your vocalist sings the lines “into the darkness we FALL!“. On “into the darkness we …” he or she is captured at -10 to -8 dB. But on the word “FALL!” they get extremely excited and hit -2 dB! This will cross the -5 dB threshold and activate the compressor to pull that bit of the vocal performance down.

But how much should it bring the signal down, how fast should it act and how soon should it stop compressing after the incoming signal falls back below the Threshold? And what do we do if we’ve lost too much of the overall loudness? Read on.

Ratio

To answer the question “how much should the compressor pull the volume down when it is activated?” we take a look at the Ratio setting.

Let’s say we set the Ratio to a value of 2:1 (read as ‘two-to-one’). This means that for every 2 dB over the threshold, the compressor will pull the sound down so that it drops 1 dB.

A quick way to think of this is to take the first number in the ratio (for 2:1 it is 2), then use that number to divide whatever the amount is that the sound exceeds the Threshold by.

As an example, let’s say we set the Ratio at 2:1. Our Threshold is set at -10 dB. A vocal line peaks at -4 dB which goes over the Threshold by 6 dB. So, with a 2:1 Ratio, the sound is brought down to 3 dB (6 divided by 2).

Another example – we set the Ratio to 4:1. A sound goes over by 12 dB (whoa!). Just take 12 dB divided by 4 dB = 3 dB! Instead of the sound exploding at 12 dB over your Threshold, the magic compressor robot pulls the volume down so that it only peaks at 3 dB over threshold, thanks to your 4:1 Ratio setting.

Side Note – if you set a Ratio of 1:1, guess what happens? Nothing! Any number divided by 1 is that number. So if your sound breaks the Threshold by 8 dB, what is 8 divided by 1? It’s simply 8, so there is no volume reduction. Engineers call this unity gain. On the other hand, what if you set the Ratio to ∞:1 (infinity-to-one)? Well, anything that passes the Threshold is brought down to the Threshold itself. In other words, the magic compressor robot has been given infinite power to totally prevent all sound from crossing the Threshold, which is called Limiting (and the topic of a future post).

The volume level that the compressor pulls the sound down to is called the Output, and a fancy word people use when they talk about compressors “pulling volume down” is attenuation. More commonly, however, is the term gain reduction to refer to the compressor’s suppression of volume.

Re-read this entire section a few times. This is one of those instances where that inverted, negative value dB metering scale gets annoying. You’ll get it!

OK, now our magic compressor robot now knows how much to pull the volume down if the threshold is crossed, but how fast should he do it? And how long should he keep the volume suppressed once the sound is no longer crossing the threshold?

Attack and Release

To really understand what the Attack and Release knobs do, it helps to understand the two major parts of a recorded sound — the transient and the decay (some people call decay the ‘sustain’ ,’tail’ or ‘body’).

A snare hit, for instance, may have a waveform (the visual representation your DAW gives to the sound) that looks something like this:

trans-vs-decay
Figure 3. A waveform of a snare hit. The transient is highlighted in red, and the decay is highlighted in blue. These aren’t “exact” boundaries, by the way!

That leading “spiky portion that pokes up real high” is called the transient. For the snare, this peak may represent that initial loud, woody crack that gives the instrument its defining character.

A bunch of information comes after that transient, called the decay, which is most likely some resonance from the body of the snare after it is struck, the residual “ringing” from the hit, and maybe some vibrations.

If something like this transient comes along and gets louder than our threshold is set, the magic compressor robot springs into action and attacks this signal by pulling down the volume (reducing the gain) to a level based on the ratio setting. How fast he responds to this threshold-breaking rogue depends on the Attack time we set on the compressor.

Once the loud, trouble-maker is gone and the input sound begins to go back below the threshold, the magic compressor robot is cued to stop pulling the volume down, or to release the compression. How soon after the incoming signal stops breaking the threshold should the volume reduction stop? Well, that depends on the Release time set.

Below is a simple graph showing how Attack and Release times relate to the incoming sound (Input). Study this graph carefully, and don’t be put off by the rigid-looking, engineering aesthetics of it, it is quite useful.

File:Audio Compression Attack and Release.svg
Figure 4. A useful graph showing the phases of attack and release time as they relate to an input signal (a sound) that is breaking the threshold (the horizontal dashed line). The red dotted line shows the output level that the compressor reduces the original signal to based on how it is set. By Iainf [Public domain], from Wikimedia Commons
We could use a little more intuition when it comes time to turn those Attack and Release knobs, so perhaps it would help to have some sort of concept in the back of our minds to latch onto so we know (kind of) what we’re doing.

For this I came up with some illustrations that represent what I see in my mind’s eye when I think of compression Attack and Release.

Check out the figures below. You should think of the red curve I’ve drawn over each waveform as the activity of the compressor over time.

Based on the beginning and ending slope of the red curve, you can kind of visually think about how Attack time (beginning-half/ascent of the curve) and Release time (ending-half/descent of the curve) relate to the audio and its transient.

A fast attack will quickly respond to the transient and pull the volume down. I’m not sure what other engineers say, but I think of it as softening the punch when the compressor attacks fast enough to pull the volume down on a portion of the transient.


study-1355437_1280NERD DISCLAIMER

The figures below are in no way mathematically significant or accurate. These are simple visual aids to understand how the compressor acts on a signal.

 


 

fast-att-fast-rel.png
Figure 5. Fast-Attack, Fast-Release. The Attack time is fast enough here to attack part of the transient (softening the punch), and the Release is fast enough to not affect too much of the decay.
fast-att-slow-rel
Figure 6. Fast-Attack, Slow-Release. The Attack time is fast enough here to attack part of the transient (softening the punch), but the Release is slow enough to suppress some of the decay.
slow-att-fast-rel
Figure 7. Slow-Attack, Fast-Release. The Attack time is slow enough here to let the transient slip through unaffected, while the Release is fast and does not suppress much of the decay.
slow-att-slow-rel
Figure 8. Slow-Attack, Slow-Release. The Attack time is slow and does not affect the back end of the transient. The Release time is also slow and suppresses a good portion of the decay.

Keep in mind that these Attack and Release times are on the order of 1/1000ths of a second! You’ll see the Attack and Release values on your compressor ranging from something like 0.0 – 1000+ ms (milliseconds).

Attacking part of that initial transient and softening the punch may or may not be desirable. 

Have a listen to the audio clip below of drums from a metal song I’m mixing right now—the compressor on the kick drum track is attacking very fast, softening the punch of the kick, making it a little flubby sounding:

 

Compare that to this next audio clip of the same drums with the attack time on the kick much slower so that the punch is restored:

 

If you can’t hear it right away, keep listening and A/B’ing the two, The effect is a little subtle but makes a big difference in how the drums “feel” in terms of impact.

For a snare hit, a fast attack that softens the punch would most likely make your snare sound flat, rubbery and unable to pierce through a thick metal mix, for example. On the other hand, such a sound may work on a layered snare in a dubstep production. Experiment!

Softening the punch in a vocal, especially on a big, soaring chorus, is something I like to do to help the vocals flow and take flight. It also can reduce some plosives (those loud and nasty “p”, “b” type sounds) and keep them from clipping.

Just as important as listening to the effect of Attack time on a transient, is listening for the effect Release has on the rest of the sound (the decay).

If, for instance, you have a track of double kick drums playing 16th notes at high speed, you’d want an Attack/Release setting that acts just slowly enough to not soften the “clickiness” or “beater sound” that makes double kick so delicious, but releases quickly enough before the next kick hits. This would avoid a “smearing” or bottoming out of the double kick run.

Knee

Knee refers to how gradually the volume is pulled down once attack begins.

Most compressors are armed with a “hard-knee” setting by default. But what does that mean?

Let’s use our little robot analogy again.

Should the magic compressor robot slam the volume down abruptly (hard knee), or gently and smoothly (soft knee) once he has entered “attack mode”? The answer to that depends on what you are compressing!

An instrument or performance that is meant to flow would most likely benefit from a softer knee setting. A percussive, fast, aggressive performance would most likely maintain or enhance its edge using a hard-knee setting.

With all things production, you’ll just have to experiment to find which setting fits that particular performance and instrument!

Make-up Gain

With a high-enough ratio and the compressor going to work, you may notice that the overall loudness of the track is lowered. This is where make-up gain comes in.

Just increase the make-up gain value (carefully!) so that you recover some of the overall loudness, while still maintaining the compressor settings.

Don’t over-do this, or rely on this to greatly enhance the volume of an already weak signal (recorded too low) as it could become harsh-sounding very quickly. Strive to use this only to recover what is lost from compression in terms of the perceived loudness.

Peak/RMS

So far in this article everything I have told you about the compressor relates to the device acting on the “peaks” of your signal.

The example above of the transient is a peak (the loudest, initial peak that would typically get out of control in a mix).

Most compressors allow you to toggle between “PEAK/RMS”. You’ll probably see a little tick-box or a switch that lets you select one or the other. By default, a compressor acts on peak values (like our transient).

You can set the compressor to ignore the peaks and monitor the average volume of the signal. A value called RMS (root mean square) can be calculated by the compressor as it receives the audio signal.

In a basic sense, RMS is “roughly how loud the signal is on average”. For a future post, I will talk more about RMS and how it relates to your final mix and mastering.

RMS mode is good for when you want to use the compressor to “color” or “thicken” an instrument or performance, but the audio doesn’t necessarily have a lot of wild volume peaks or transients. Something like a fuzz bass, a synth pad, soft vocals, etc would be potential candidates for trying out the RMS mode of a compressor.

‘Hearing’ the Compressor Working

This was the biggest struggle I faced when learning about compression.

I would make extreme tweaks to Attack, Release, or Ratio and I couldn’t quite tell if I really heard the sound changing or if I was imagining it!

A big change in overall volume, or the perceived loudness, is obvious, but I could not hear the Attack and Release setting changing anything at all.

The trick to is to feel how the sound is being affected. When you are adjusting Attack and Release on a bass track, for instance, try to feel the “punch” of the bass changing. Try to feel its stability changing. Crank your monitors and see how it thumps and resonates in your body as it plays. This feeling for “punch” is good for the kick as well.

Setting an extreme Ratio setting (20:1 or even 100:1) and a low Threshold (so that most of the signal is being compressed) can also help you hear what Attack and Release are doing. Just don’t forget to return the Ratio and Threshold settings back to a sensible value!

Feeling for the subtle changes offered by the Attack, Release, and Ratio settings takes a lot of practice!

Additionally, you always want to listen to the compressor at work on a track in the context of the whole mix. If certain parts of your guitar solo, for instance, kind of dip down and disappear beneath the rhythm section, perhaps the attack time is too fast and the initial transient made by the guitarist picking the note has softened too much. Or maybe the release time is too slow and too much of the sustain of a note is being compressed after the transient triggers the attack.

Conversely, if you’ve compressed a vocal track and all you can hear is plosives (hard consonants), then the Attack time may be too slow, and softening the transient with a faster Attack may allow you to raise the overall level of the vocal so that it is even and smooth, and most importantly always audible.

Creative Compression

As I said earlier, and I’ll keep saying, there are no concrete rules to audio production. Experimentation can give birth to fresh new sounds and even entire musical movements.

The characteristic “pumping” sound of techno and other electronica-style drums comes from using the compressor “the wrong way”; some producer/artist back in the early 90’s or so must’ve wanted to juice up a drum sequence and went crazy on the Attack and Release knobs until this exciting, breathing “pump” energized the track.

This pumping, which I’ll go into detail in a near-future post, is also really useful on your parallel drum buss to give more energy and “slam” to the drum track.

Other producers will use compression in countless other creative ways to color or shape the sound of an instrument.

Screen Shot 2018-05-24 at 4.08.19 PM.png
Figure 9. Stillwell Audio’s incredible Rocket Compressor. This compressor is great for general use as well as excellent for coloring, experimentation, and getting your pump on!

The More You Know!

Having the fundamental understanding of the compressor is the first step in informed experimentation and honing your own sound. This initial learning curve is the easy part, now you’ve got to practice, practice, practice and keep going.

Remember, not all tracks need compression, others are in serious need of sonic discipline! At other times, the compressor can act as an FX unit to achieve something novel and interesting.


Let me know in the comments section if this article has been helpful to you. Feel free to share it, and message me if you have any burning questions.

Stay tuned for more posts like this (and beyond)!

2000px-F_icon.svg.png  Like and Follow on Facebook!

 

6 Comments Add yours

  1. Ernesto says:

    Hello Keith,

    Thank you for the thorough explanation. I am fairly new at mixing and your post made me understand what the terminology means.

    Have a great weekend!

    Ernesto

    Like

    1. Keith Prater says:

      Hi Ernesto,

      I’m very glad you found the article useful! Thanks for following, there’s more good stuff on the way.

      Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s