Sounds cool. So, ready to talk some shop?
I'll just lay some groundwork first. It'll be like a lecture.
Split-Spectrum Compression
Most of the music I'm working with is made by vibrating membranes — be it a metal string or a vocal chord or a drum head, it's a vibrating membrane of some sort. And a universal characteristic of vibrating membranes is that they vibrate more strongly at some frequency ranges than others and that intensifies as you play louder, which, in effect, means that any instrument you play louder tends to gets more midrangey.
Because the upper harmonics are coming out more?
Just because some frequency ranges are easier to reproduce for a vibrating membrane. You know this intuitively if you play a bass string really lightly. It has this huge, fat sound, but then if you lay into it, it sounds like "clank, clank". When somebody's talking really quietly, you can hear all of these highs and lows, and then if they start yelling, it becomes more midrangey. To some extent, this is present in every instrument. You know how drums being played hard sound when you hear them through a mic? It's always struck me as not quite right. It's not what you want to hear. When things get louder you don't want more midrange, you want more of everything. The idea, therefore, being to compress the midrange frequencies more than you compress the highs and lows so that, as an instrument gets louder, the spectral balance remains pleasing. You can really just divide a lot of instruments into highs and lows then compress the highs one way and the lows another way. It's not always the mids that pop out, sometimes it's the highs and the mids. It's not universal. A typical application is a bass guitar. Every time a player goes up to the high strings, the bottom drops out, and every time they hit the low E, the thing just gets outrageously fat. It's annoying! You can't get a consistent low-end picture. So in my studio I have a crossover, which is like what you'd use for a public address system to separate the lows, mids and highs. I can either use it as a mono three-way or a stereo two-way. So the first thing I do when the bass leaves the tape deck is run it through the crossover, and then I run each of those signals, the highs and the lows, through their own compressor. Then I bring those up on the board so I have a high-end portion and a low-end portion. The advantage of using a crossover is that when they recombine, there's minimal overlap.
You don't have phase problems?
You have less phase problems.
So your crossover is pretty discerning?
I have one that has a pretty high slope≤24 dB per octave, which means that one octave away from the crossover frequency the signal is attenuated by 24 dB In a normal mix this is inaudible. Some times a gentler slope is preferable. I have one of those too. But then you've got to choose the right compressor for the high-end and the low-end. Typically you want different compressors. The one that you're going to compress the low-end with has got to have much slower attack and release characteristics and be something that sounds punchy with low-end. Whereas the one for the high-end wants to have a lot faster response. I like the Urei 1178 for the low-end. But there are many that will work. You've just got to know your collection of compressors to make the right choice. Now let me give you another example of the power of this technique.
You sound like a barker.
(Laughs) It's the gong/kick drum sound, which you get if you put a nice coated Ambassador head on a kick drum, tune it so it rings and maybe play it with a mallet, sort of like a timpani. You know, you're doing a marching band kind of thing where you want a big "BOOM, BOOM, BOOM". But if you just put that in the mix and say, "I want some low-end on that thing," it just goes "wooom, wooom". It just clutters up your mix. So what you do is send it through a crossover. And then, because it's a drum, you gate it so that the high-end portion and the low-end portion have different release times. Then you compress them also, of course, to keep things in control. But you gate the low-end portion with a fast release time so the low-end is now going "wup, wup". But you leave the high-end gate open for a long time, maybe you don't even gate the high-end. Maybe you compress the hell out of it with a really fast-release compressor so that it accentuates the "kaaahhh" sound.
The tail of it?
The tail of it. If the compressor's release is really fast, it goes in and follows the tail down. In doing that you can get all of this personality, which is, in a drum like that, above 500 Hz. The "boom" is down below 200 Hz. I often end up crossing over around 200 Hz. Sometimes what I'll do with a bass guitar is record through an amp that is a little overdriven and fuzzy sounding, without much low-end, and also record a direct signal. Then, instead of taking just one of those and crossing it over, I'll take them both, cross them over at exactly the same point and use just the low-end from the direct signal and just the high-end from the amp. That way, not only do you have individual controls over the lows and highs, but the lows are clean and punchy and the highs are gritty and a little distorted, which makes them more interesting under a lot of conditions.
It gives you a lot more information up there.
Yeah, information! Like on the Nine Wood CD (two basses, drums and a vocal), it was crucial. On my first attempt to record those guys, I wanted high-end in the mix, but the high-end on the basses was all "ping, ping, ping". I wanted some "szhhh". So I tried a different approach and said, "Okay, you're going to play through this amp, and I'm going to turn it up so it's gritting out and then I'll record that on a track. I'm also going to have a direct box and a track where I'm recording your regular amps with more lower mids, so I'll have a few different degrees of edginess and distortedness and high-endiness. Then I'll recombine those as I mix it to get a nice full high-end, which is the hardest thing to get without a guitar." I realized that I sort of count on guitars to fill out the high-end of the spectrum. So with a band like Nine Wood, I needed to treat the basses more like guitars.
Do you ever do this with vocals?
No, I never do. I never have. I might some day. Vocals I consider much more of a voodoo area of recording. Basically, I've got my Neumann U87 and I've got my Urei 1176 and an old east-German tube mic, and, you know, you put them up and you hope for the best. You can reposition the mic, but if you try adding low-end with EQ, it can sound really bad. But that brings me to another thing that I do. It's called frequency-dependent compression or dynamic notch-filtering.
Frequency-Dependent Compression
There are only a couple of units on the market that do this, and they're not too popular because they're kind of complicated and confusing to use. The ones I have are the Urei LA22 and the Brook Sirens DPR901. These are compressors that will compress just a narrow band of frequencies. For instance, if a vocal is going to get piercing, it's usually going to be around 2.5K, 3K. So if you've got a problem, you just set this compressor for that little range, and then every time the vocalist honks in that range, the honk and only the honk is pulled back. This allows you to really crank up the high-end on a vocal and get this really shimmery present tone, which is nice, but without getting so much piercing midrange.
Can you dial in more than one frequency?
With the Brook Sirens you can. It's got four separate bands, each with its own frequency, threshold and bandwidth. I usually use one band around 3K and another, to de-ess, at 9 or 10K. The only thing I don't like about the Brook Sirens is that it's a little slow. It's an excellent vocal box. If I want to do this type of compression with instruments, instead of vocals, I'll usually go for the LA22 because it's faster. But with the LA22, each channel just does one frequency. These are units with two channels, one frequency each. They're great for toms and kick drums if the player is using too much dynamic range. You can do normal compression, but you can't keep the hard hits from being really attack heavy. You've experienced this probably.
Oh, yeah!
You can compress the hard kick drum hits until they stay in one place, but then they go "ATTACK", and the soft ones are just "thump, thump" and you're going, "Jesus, this is frustrating!" So you can use the crossover technique to control the high and low-end independently. What I often do now is compress the whole kick drum and then run it through an LA22 just compressing the area from around 1K to 4K or 1K to 8K, all the attack areas. I hold those in place independent of holding the whole thing in place. Then if I want more high-end on the kick drum, I move the threshold up and if I want a little less, I move the threshold down. It really allows me to sit the kick drum right back in the mix. If you don't have one of these esoteric boxes, you can compress your kick drum as usual, then cross it over at, say, 500 Hz, then run the highs through a second compressor. Since we're trying to affect the attack, we're going to need a very fast attack time. A much slower release time will help you avoid unwanted clatter. A SpectraSonics 610 would be my choice.
So are you doing a lot of this application during tracking?
No. This is all done during mixing.
Couldn't you do a light application while tracking, just to get your levels to the deck a little hotter.
That's true, I could. What do I do when I'm tracking? I always compress vocals. I always run them through the 1176, usually slow-attack, fast-release. What else do I compress to tape? I'll do some high-end percussion, like shakers and tambourine stuff, again with the SS610. I'll compress those just to tame the transients a little. For the most, part though, I don't compress things to tape. I just try to be careful to get the right levels. It always makes me uncomfortable to do too much as you're going down to tape because you're never quite sure what the mix is going to turn out like. So as much as I can, I focus on getting good sounds and then leave the rest of it for the mix.
That reminds me of some technical liner notes I just saw. They listed a couple of microphones, and then they asserted, "Individual microphone equalization is not permitted!"
There's this album by Jerry Garcia and David Grisman where they did the same thing, there was no EQ, just the mic. And of course, the Shellac (engineer Steve Albini's band) albums. He does all this stuff where he records right onto lacquer or whatever, the whole band with only four mics. But he does incredibly minimal processing.
He's also got a great collection of microphones.
He's got incredible stuff. And let's not forget that he's using certain kinds of natural compression. He's using old tube preamps that will compress a little bit, and he's using guitar and bass sounds that are distorted Distortion is a kind of compression. So he says he doesn't use any compression, but it's sort of a question of semantics.
Yeah. Well I was just reading this article about getting, "in your face" mixes. They started off by knocking compression, but then they went on to say how much they use it in all these different ways. I think they were trying to warn against over doing it, because there is so much squashed stuff out there, but it came off kind of funny.
Would you say that my stuff sounds "in your face"?
Oh, yeah.
