fbpx

Compression: Yes, You Can!

Picture 1.png

So I think I finally stumbled on a way I’m really digging to teach people how to learn to use a compressor. There’s a ton of info out there on what the controls all do followed by what someone thinks are where you should set the thing to get started in either specific numbers or some sort of vague fast/slow/medium terms. This stuff is fine and good to check out to get started, but in my opinion these training/explaining approaches generally fail at teaching engineers what to really listen for as well as a systematic approach to setting a compressor. Last time I checked it wasn’t about numbers, it’s about music and how it sounds. And judging by a lot of what is heard these days, I really wonder if a lot of guys that wax poetic about compressors really hear what the things are doing. A big danger with compressors is that they ALWAYS raise the average level of an input; it might be quieter and the signal level might appear lower, but the AVERAGE level is higher meaning the amount of dynamic range between the highest and lowest level is reduced. With that higher average level, the input often gets turned up and things that are louder can trick our ears into thinking they’re better if we’re not careful. I think this approach can also help with hearing past the veil of louder.

As is typically the case for my methods, I have been enlightened by another engineer. This is mostly adapted/taken from a book I recently finished called Mixing With Your Mind by Michael Paul Stavrou aka Stav. Stav’s name might not sound familiar, but he’s been around a bit. He started out at Air Studios over in London which is George Martin’s place and has worked with some people you might have heard of like Paul McCartney, Elton John, Cat Stevens, the Pretenders, John Williams(as in movie soundtracks), and Roberta Flack. Stav is a studio guy, but at the end of the day mixing is mixing and he’s got some great stuff in the book. It’s pricey, but I’ve become a big fan and always have my copy with me. If you read this approach and try it and dig it, please go order his book because most of the credit really belongs to him and there’s more cool and practical stuff in the book; I’m digging his approach to figuring out what all those different settings on your reverb do. Anyway, here’s how I’m training folks in setting compressors when I’m working one-on-one with them.

***

We need to start by configuring the compressor’s controls for our purposes. This might seem counterintuitive in a way because we’re going to do a number on our sound, but hang with me. First set the release and attack as fast as they go. Next set the ratio pretty high. You don’t need to set it up as a limiter with an oblivion:1 ration, but 8:1-12:1 might be a good place to start; Stav says to set it as high as you can go, but I don’t think that’s necessary. Now drop the threshold or drive the input so the thing is really workin’. We don’t want 2-3 dB of compression. We want lotsa compression and pumping nastiness with the meters bouncing around quite a bit. Seriously, make sure you screw the thing up. Add any makeup gain you need so that you can hear your instrument now that you’ve brutally compressed the input. Preferably add your gain using your channel fader in lieu of a makeup gain control on the unit; this is just a preference as I find it easier to pull the fader down later when I’m fixing the messed up sound and the level comes back up. The compressor will pump and probably sound pretty messed up, BUT you will hear the unit which is great for folks new to this. If it’s just a mess of distortion, you might need to adjust the release so it’s a bit longer but get it pumping like mad.

So here is a great approach for adjusting the controls once you’ve set it as above. It’s all about the order you adjust things, and following this order will help you really learn to hear what each control does. Here’s a simple mnemonic for remembering the order: Setting a compressor is an ART.

Now remember that word “art” and add an extra “r” in the middle so we have A R R T. Here’s what it stands for:

1. Attack
2. Release
3. Ratio
4. Threshold (or input)

1.) Attack
Start by adjusting the Attack control; for me, this is where the money is in this approach. You don’t need to start at the fastest setting, but it’s my personal preference because it’s where you’re inflicting maximum damage that we will heroically heal through our adjustments. Slowly adjust the length of the attack listening to the leading edge of the instrument while trying to ignore the pumping. The Attack knob can work almost like a tone knob. Faster attack is going to give you a “thinner” sound and slower attack will give you a “fatter” sound.

If you want the science of the thing, Scovi taught us about the 1/4 wave concept at his seminar last summer. You need an attack time 1/4 the wavelength of the highest frequency you want to affect. Higher frequencies have shorter wavelengths and remember with sound that distance is also time. Shorter distance = shorter time so higher frequencies need a faster attack time to be affected. You can almost think of the attack knob as a high-pass filter for your compressor. A faster attack will let higher frequencies pass unaffected by the compressor. So to control overall dynamics, you’re probably going to want a faster attack to make sure your unit reacts in time to affect a larger portion of the entire frequency spectrum of your sound. But if you want to shape the sound–yes, please–you’ll maybe want to play here a bit. Two quick examples with vague, non-commital speed settings. The fastest attack on an acoustic guitar might really bring out the sound of the pick on the strings, but slowing the attack could de-brutalize a scratchy acoustic. Slowing the attack on a snare drum can add body to the snare, but speeding it up can enhance the crack.

2.) Release
Stav looks at release as a rhythmic component using the release to give the instrument a subliminal groove. In the world of sound reinforcement, I’m not sure how feasible it is to always think this way since you’re most likely not going to have the ability to adjust your release for every song and even with digital consoles I haven’t gone that far…yet. I like to simply try and adjust the release to get rid of any pumping and lately I’ve generally been trying to go as slow as I can so that the comp is either completely or almost completely released before the next time I want it to kick in. I like to use a more faster tune in the set to adjust this otherwise things can get wacky on those fast songs when the comp isn’t releasing as fast as you want it.

Here’s another way I think about this, though. When I want to decrease the dynamics of an out of control instrument to bring it under control, I want to go faster on the release as long as it sounds natural without weird pumping artifacts or distortion. For example, with an amateur vocalist I want to turn them down when they get out of control loud but release smoothly enough and fast enough so their level is back up in time for the quiet stuff which could be within a line depending on the vocalist and material. To even a bass guitar I want the release to be as even as possible to give the effect of adding sustain by gradually raising the volume of the note as its amplitude dies off. On the other hand when I’m compressing a snare drum, I’m working on shaping the transient so I try to go as slow as possible to eliminate ring and other garbage hanging out between snare hits without losing cool ghost note detail stuff.

Don’t be afraid if your release control looks like it’s set pretty slow; ignore the numbers and position. In fact, the whole idea of this approach to setting controls is to NOT worry about how things look and to focus on how things SOUND so don’t even look at the knobs while you’re adjusting them. I think part of the idea behind using a dynamics processor is to replicate a natural fader move, and your compressor most likely can move much faster than your finger on a fader can. So in other words, slower release times can sound more natural and shouldn’t be feared. But faster ones are cool sometimes, too….

3.) Ratio
Ratio is your “size” knob, but it can also be your “control” knob. The higher your ratio, the smaller your sound. A lower ratio is going to give you a bigger and more open sound. Conversely, a higher ratio will increase control and a lower ratio will decrease control.

Adjust the Ratio to find your preferred balance of size and control. I’m pretty spoiled right now with musicians that generally possess dynamic control so I really like to think in terms of “size” more than control. And when you start thinking in terms of size, you can really start getting creative. For example, do you want your BGVs to be as big as your lead vocal? Maybe these compressor things can help “shrink” something into its place in the mix…hmmmmm….

4.) Threshold
Now raise the threshold or turn down the input so you’re not killing it. Also go ahead and put your fader back where you like it and add any makeup gain on the compressor if you feel like you need it.

These days I like to get it so that the unit isn’t always compressing. BUT that’s not a rule. I recently had the privilege to sit behind a great FOH guy mixing a pretty big artist with a killer mix and some of his inputs were always compressing. So forget about rules and numbers and what meters look like and use your ears. Trust your ears. Don’t be afraid to get this wrong. Record your mixes and listen back six days or weeks or even months later when you’ve long forgotten what you did. The stuff you hate will stick out at you, and you’ll probably be more in tune with that stuff the next time you mix.

Now one last note for those of you with a Knee control on your compressor(s). I’m not really sure where the best place in this sequence is to adjust the knee. I can see arguments for putting it after just about any step, although it’s probably good to do it before your final threshold adjustment. I also can’t give you a definitive, “listen for this” type of thing because for me it’s just sort of a sense of what I prefer on an input. Here’s an example. With the advent of Venue, I find myself trying different types of comps on things more often because I have more options available. What I like or don’t like about a particular comp is probably largely “knee” related. So if you have a Knee control and something doesn’t seem right, maybe try adjusting the knee. A softer knee is going to be more “transparent” and a harder knee is going to be more aggressive sounding meaning you will hear more of the compression effect.

***

So over the years I had sort of developed a mish-mash of what Stav suggests doing, but over the last few months I’ve been more closely following his systematic approach and showing it to our volunteers. A couple benefits I’ve found from doing things this way is that I am generally more satisfied with where I tweak things and find I can set a compressor even faster than before. I am also getting more and more satisfied with what I do when I tweak things out of order because my ears are even more in-tune with what to listen to while I’m adjusting a particular control. However, the biggest benefit I’m seeing is that our volunteers’ eyes get really big when I start doing this on a snare drum as they start to really understand what this compression stuff does.

So try this stuff out for yourself if this is new to you. Try showing it to your volunteers. If it works and your mixes improve, please go pick up Stav’s book.

David Stagl

12 Responses to “Compression: Yes, You Can!

  • Justin Whidden
    16 years ago

    Great Post! That was extremely useful and helpful information. Keep those kinds of posts coming.

  • Dave,

    Thanks for the mic info on Reid’s cymbal post. Also great post on compression. Thinking “size” with the ratio! I posted this kind of info for our volunteers but I think I might amend a bit, the ART approach makes alot of sense.-d

  • From someone that is trying hard to learn as much as I can, you have some great post and I thank you for that. I look forward to reading more and someday in the future meeting some of you guys at a conference. It the meantime I’ll keep trying to improve and difinitely continue reading blogs! Thanks again Dave and you are certainly “Going to 11”.

  • Great post. I’d like to hear your thoughts on using compressors without Attack and Release functions like some dbx products and El-Op compressors. Again, good post.

  • It really depends on the compressor. An LA-2A is a great compressor/limiter that doesn’t have attack and release controls. Same with the Manley ELOP. Some of the dbx and Drawmer stuff, on the other hand, I can typically do without. I have a bunch of Drawmer comps in our analog room that are auto attack/release and they drive me nuts because I have critical inputs I would rather have attack/release controls. But I guess the auto units aren’t the end of the world, and sometimes it is better to have one of these than nothing. BUT, with a critical input like a lead vocal, I might opt for no compression over a cheap auto unit.

  • I’ve been following your site for a month or two now, and I too picked up Stav’s book a couple months ago. It’s really good material, and I encourage anyone who’s serious about their mixing to pick one up. It’s well worth the hefty price tag!

  • I tried a little of this last night when I played back our rehearsal from Pro Tools. I’ve always been unsure on where to set the attack/release/knee and have just gone by what people have said I should use for different situations. This has really helped me to hear what those settings are doing and determine what sounds best instead of just going by the numbers. Until last night I had only a fuzzy understanding of the knee control, but I saw and heard what it was doing and the light went on in my head. Cool!

  • Hey Dave, great article. Another tip for setting threshold is to set it during the Chorus of the loudest song. If you set it during a not so loud part you will most likely end up squashing your signal in the loudest part.

  • Nice article, Dave. Like you, I’ve been doing a version of this for years, but it’s great to have an organized process to follow. This will make it so much easier to teach volunteers (as opposed to, “Here’s the ratio control…hear that?” method).

    I’ve also found that playing with various compressor settings as I post our sermon recordings in my DAW gives me a great chance to listen to the effect of all the controls in a safer, non-destructive setting. With some of our speakers, I’ll get into crazy compression because the just sound better that way (for the podcast anyway). And you’re so right, it doesn’t matter what the numbers are, just that it sounds good!

  • Phillip Graham
    16 years ago

    Hey Dave, greetings from Candler Park. I want to take a little bit of issue with the thought of long release times, especially with the current feel of record production, and the practical realities of live sound.

    Much like Mr Scovill talks about the attack time, release times can be thought of as defining a threshold frequency where harmonic and IM distortion begin to kick in. Practically this means that instruments with less low frequency content can get away with shorter release times. And this is often advantageous to a certain sound.

    A great example of this in practice is the Distressor, which is very similar to the 1176 from a gain reduction circuit perspective, but has a wider envelope of available attack and release times (slower attacks, and shorter releases).

    In many situations in live sound I find myself wanting two compressors, one to catch peaks, and another to shape the tonality. The BSS analog comps have a nice combination of these features in one unit, and I am sure there are a bunch of plugins that do, too.

    Short releases can be musically interesting on percussion instruments, sometimes the trash that the short release generates makes for an interesting grit. I find, especially on vocals, that if the release time is too long, the presentation becomes “shouty”. On drums a longish attack (25-100ms) and the absolute shortest release is musically interesting. In general I find releases between 60ms and 350ms the most useful. The amount of low frequency content in the source is ultimately going to dictate how short a release you can use.

  • I’m with you Phil. I guess this sort of points out why I hate using the words “long” and “short” when talking about compressor stuff; they’re too relative. I guess when I talk “long” I mean relative to the attack time because my release times are generally much longer than the attack time. The times you’re giving are probably pretty close to where I actually land, but I really just try and listen and put it where it sounds right to me.

  • Phillip Graham
    16 years ago

    I honestly never thought about the relationship of attack time relative to release time, in terms of long/short. That is a good point!

    60ms is really darned short, but it can work great on things like vocals, or snare. I would actually consider going shorter, but good luck finding gear (other than dedicated hard limiters) with shorter release times. On snare in particular the short release leaves this trash in the snare decay that has become quite fashionable on pop/punk records. The problem is when such things bleed into the other sounds in the mix, which often happens at mastering. I don’t want all that snare IM showing up in the vocal line on top of the beat.

    Once you learn to listen for this release trash on the decay tails of any transient (vocal or percussion), you will throw out pretty much every CD made after 1995. Its really bad when the snare decay IM “warbles up” the cymbals in the OHs. I have noticed that the folks that make mixes that survive really severe limiting in mastering (e.g. Chris Lord Alge) have as little overheads as they can. A currently popular and egregious example of this is the CD “Riot” by Paramore. Listen for the “frying bacon” at the back end of each snare hit, it will quickly drive you crazy!

    Here is a personal rule of thumb that I use, mostly when setting speaker limiter decay constants as a systems tech. It should apply equally as well as a guideline for setting compressors when mixing: The release time needs to be ten-fifteen times the length of one period of the lowest frequency in the bandpass. So for a box that is high passed at 100hz, that means a limiter release time between 100-150ms (adjust to taste by ear). I’d say that this rule holds up to about 1khz. Above 1khz the release time usually needs tailored to be flattering to the volume envelope of the types of signals produced. This can mean much longer release times in the highest octaves (200ms or more).