Guest post by audio engineer Todd Watson

PART I: THEORY

When learning the basics of recording, the subject of compression often seems daunting. When to use it? How much is too much? Too little? What ratios and attack times are best? Why compress at all? Or maybe none of the above, you may have a life. In the following article, I’ll try to shed some light on the issue without becoming overly technical, and then focus on real-world examples of how to enhance your sources with the right compression.

First, some history: Compression was developed by radio broadcast engineers in the 1920’s to automatically detect and lower peaks in the program material to “level” or “compress” the transmitted AM signal, preventing distortion. This is why many early (and current) compressors have level somewhere in their name, and those ugly, ugly industrial paint jobs. Soon after, duplication engineers began using them to do the same thing when cutting master discs for wax, then vinyl disc duplication. If a recording was too loud, had too much bass content or the transient peaks were too sharp, the cutting lathe would literally jump out of the groove, ruining the lacquer master and interrupting drinking. Adding hardware compression and equalizers before “cutting” the record made the process much easier and had the added bonus of making recordings more consistent at lower volumes on a wider range of speakers.

Today, mastering is more common than ever, with nearly every recording currently released eq’ed and compressed or “leveled” in some way. Soon, compression moved to individual instruments, especially when close instrument and vocal mic placement became the norm in the 60’s and 70’s. New instruments with a wide dynamic range like electric guitar/bass and the modern drum set would now typically be compressed, and a more consistent volume for each note could be put to tape, making mixing easier while doing drugs.

On vocals, singers with softer voices could be compressed to help every syllable cut through a full mix, signaling the end of big voiced singers until Meatloaf. He actually sang directly to tape. No mic, nothing. Then Bono Highlander-ed him. There can be only one… Different types of hardware compressors were developed, each with its own character, each loved and hated like Pee Wee Herman. Software compressors have evolved frightening quickly to faithfully model classic analog units, sometimes combining the best elements of several designs or doing tricks impossible with hardware.

But I know what you want to know, how do it work?

PART 2: CONTROLS

Most compressors, either hardware or software, have a range of controls that look like Stephen Hawking’s ipod, but in practice, are really simple. Starting with what they do in the signal chain, here they are:

1. Threshold is the point at which the compressor starts to “hear” the input signal and process it. Think of this as your “zero” point. Use the attenuation/reduction meter to help set the compression level. If you see large amounts of compression on your input signal (more than 5db), raise the threshold. If you see little or no compression, lower it until you’re compressing 0-3db on average. This is a subtle tracking setting, mixes may require more, but don’t they always?

2. Ratio is the amount that the signal is reduced (or compressed) based on the strength of the incoming signal. 2:1 compression will compress -2db for each 2db above your threshold point and output only +1 db. 5:1 will compress -5db and output +1db and so on. The first number is the db you want reduced, the second is the 1db of output after compression.
Anything over 8db is generally considered limiting, a process which sets the absolute level of a signal through fast, steep, devil-may-care compression and absolutely precise output control.

3. The Attack and Release knobs determine how quickly the compressor starts and stops affecting the input signal. Auto, Fast and Slow are all curves determining how quickly the compressor reacts to the source. Auto averages transient information it detects from the incoming signal and sets the compressor attack and release “Auto”matically. It is a GOOD THING. “Pumping” and “Breathing” occur when Attack and Release don’t match the transient speed of the input source or when you visit the wrong rest stop.

4. Knee or Soft Knee is a term for a softer sloping compression curve, (like the bend in your knee) that is smoother on attack and release than the more sliding-board curve of the standard settings. It may work with or independently of the main Attack, Release or Auto controls. Hell if I know, it’s your compressor! Read the manual! God!

5. Level, Gain, Make-Up Gain or Output is essentially a volume control which lets you manually set the final output of the compressor, adding or subtracting volume to the processed sound. Use this to set the level to your recorder or mixer fader. Leave 3-5db of headroom for later processing or fader rides during mixing.

There are many types of hardware and software compressors and they all have different strengths and weaknesses, like a ticklish weightlifter:

• Tube compressors give a warmer, smoother compression but at the expense of high-end crispness at extreme compression.
• Opto, Optical or Vactrol compressors have many of the same tonal qualities as tube compressors, but are more transparent and forgiving at high compression rates.
• Solid-State compressors are very transparent and color the signal least at low-medium reduction settings, but at extreme compression settings, the signal becomes highly colored in an unnatural way. Usually, NOT a GOOD THING.

However, unnatural sounds can be of use in some situations. Like many elements in the recording process, distinctive, individual tones can come anywhere in the signal chain. It’s all about using your wits, your ears and finding what’s appropriate to the track to use compression musically. I’ll talk and talk about how in the next segment.

PART 3: PRACTICE

Here’s some general compression tips, and then some more advanced voodoo. Most apply to both hardware and software models. Watch your ears as you experiment, compressors can turn out volumes that will rid a house of varmints for a fortnight.

Lightly compress when tracking vocals and widely dynamic sources. I know that many digital purists claim that at 24bit depth there’s now no need to worry about noise floor or bit-rate, just record lower, compress later. I recommend doing both. The sound of a lightly compressed vocal will immediately give a more produced, polished sound in the artist’s headphones as they track. And as you probably know, a confident, wow-I-sound better-that-I-thought artist is halfway there to a great performance. It also helps a tracks consistency when you mix. No more that 2-3db at a 4:1 ratio though, unless you want to color the track, and that’s HARD to get out. It’s way easier to add more compression later, when you can experiment to your hearts content with more effect-type settings without being stuck with them. Or just record one with light compression and one without to another track. With nearly unlimited DAW real estate these days, why not?

Use different compressors for what they’re best at. A guitar that has a very bright top/thin low end is a bad match for a solid-state tone. Likewise a heavy, deep bass doesn’t benefit much from the softening and warmth of a tube type. Take each sound as it comes, and see what it might benefit from. Same thing applies to vocals. Does the singer get brighter or darker as they go for it, is their tone more edgy and thin, (tube) or round and darker (solid-state)? Again, during tracking, you don’t want to go too far with changing voice timbre through compression, but in mixdown, the sky is the limit. (pun SO intended)

Two units are better than one. If you need 6db of leveling, use a few db of your most transparent compression on really dynamic sources like vocals, bass and acoustic instruments. Then use other, different flavored compressors for a few more db when mixing to get the best of both worlds, rather than a lot of one. Try making copies of the track and using different compressors with different qualities, subbing out one for another or blending them when appropriate for the song.

Do NOT compress your mix before professional mastering. Many mastering engineers, (myself included) have been receiving tracks that are limited or compressed to -.03 or OVER full-code digital signal, leaving no room for eq, multi-band processing or even a decent chance to let the track BREATHE. I know it’s sweet to throw that limiter across the master fader bus, dig in 5 or 6db and hear the magic happen. And that’s cool for making copies for your car or to groove to. But PLEASE take that stuff off before you run out your final, highest-resolution-possible master. Leave -6db for the mastering engineer to do his/her thing and you’ll get results that will make you cry with joy.

Sometimes, move the damn fader, Jim. If you find that you just don’t like any of the post-tracking compression you put on, just manually ride the fader. It’s much more transparent and leaves more of your mic/preamp tone intact, but does require having enough fader gain to work with and Spock-like patience. Make sure to also leave some headroom in your instrumental mix at the master fader for the 2-3db that vocals add. It’s totally old-school retro, man!

Ignore number four! (Sort of). I’ve sat at the elbow of well-known mixers and seen the whole mix go thru a compressor, even with NO compression happening. An $11,000 compressor, but still a compressor to get their “sound”. At a seminar with two multi-platinum producers, one swore by compressors at the master, the other swore at it. Jimmy Douglass: “I just don’t like the sound.” Brendan O’Brien :“It sounds like a RECORD!” and no, he wouldn’t say which one! If you prefer the sound of a classic limiter at the master fader when you print your final mix for mastering, lower the main master fader until you get a peak of -6db. DON’T do it at the output of the compressor or plug-in. Some of these software simulations even mimic output gain structure, so it may change your carefully crafted tone.

Compress before equalizing! By compressing before the equalizer, the eq is applied in a more consistent way across the track. If you’re trying to get more “air” into a vocal or instrument, putting the eq first will result in the air being rolled off or otherwise changed as compression kicks in. Eq after and the air stays constant across the top end. This is also true for background vocals. Send them all into an Aux Bus or sub-group, compress it first, then add eq and effects. “It sounds like a record!” Piano, horns and acoustic guitar also need compression first, to keep the eq sparkle as they get louder.

Equalize before compressing! If you’ve got tracks or mixes that sound good when things are at mid-volume and energy, but the sound gets edgy as the track gets more aggressive, use compression color to smooth it out. Got a singer whose top end is just a little harsh? Let tube or optical leveling take some top end off and add warm harmonics. As some instruments get brighter into their top range, the roundness of a tube compressor helps tuck them in, while keeping them up in the mix. Full mixes can also benefit from this combo, the compression adding a tape-like tone to thin lower mids in a way eq doesn’t. Solid-state’s tonal edginess as it gets hit harder is great for singers or sources that lose high end as they gain volume.

Sub-mix, compress and eq things to gel them together. Set up a Stereo Aux Input and send the drums to it. Or just the kick and bass guitar. Or two guitars. Or reverb. Or background vocals. Then compress and eq the living hell out of it! Then blend them back into your original tracks. I’ve seen hip-hop producers who had the same kick drum in five different DAW eq/compressor channels to make it sing in all registers. This also works great on lead vocals, adding air and bringing out every breath and luscious detail. Use a high-pass filter set around 1k after the compressor to keep things from getting too muddy. Just DON’T use it on everything!

These are just some basic ideas, and hopefully a jumping off point for your own signal chains and concepts. As you can see, there are no hard and fast rules, so don’t let compression intimidate you. It’s a powerful tool in getting a consistent, polished sound. To quote the rapper Cannibus: “I’m never under pressure, cuz I keep the pressure under me…” Feel free to leave a comment here with questions and opinions and I’ll answer them as soon as I can. Good mixing!

Todd Watson is a producer/audio engineer with twenty five years experience in nearly every aspect of audio production, including studio and location recording, mixing and mastering, as well as post-
production scoring, sound-design and audio sweeting for tv and film. His work includes artists ranging from Greg Allman to The Game, and corporate clients as diverse as The Atlanta Symphony Orchestra,
Cartoon Network and Yamaha. He is currently based in Atlanta, GA. Check out some of his tracking, mixing and mastering projects and scoring and sound design work