The Light of Day

So you want to make an album? (part 17)
KeithHandy posted in Producing, So You Want..., Tools on July 18th, 2007

Note #1: the word “band” in this installment refers not to a hodge-podge of long-haired freaks, but rather to a slice of the audible spectrum, as in “ten band equalizer”.

Note #2: I’m thinking about how to do re-do the image for this post. Obviously most of the images in this series are temporary placeholders, for copyright reasons. But this one in particular might be a real turn-off, even just for the blog. (Not to mention the faux-pas of having the salad fork on the inside.) Just be aware that I was primarily trying to merge some of the concepts in the article, using simple images, and the overall effect is a little more gruesome than I expected.

To read the entire series, go to the “So You Want…” category.

Installment 17: Dissecting the spectrum

Why did we all dissect frogs in biology class? We couldn’t have forseen it at the time, but in retrospect, it turned out to be a real advantage once the time came to build a GIANT DEADLY LASER-TONGUED ROBOT FROG to unleash on our enemies. Would you want your master plan to backfire on you just because you put the cloaca where the glottis was supposed to go? I didn’t think so.

Although it will benefit your album to conduct similar experiments on songs that already exist, the tools for dissecting music don’t work quite the same way. You can’t, say, carve out the funky rhythm guitar and set it neatly on a paper towel next to the song. Equalizers dissect the music in cross sections, as if you sliced your frog cross-ways like a salami. Some organs may be small enough to remain intact in a single slice, but larger organs will be split between two or three slices, and then things like the spine, muscles, and skin will be distributed among all the slices. (If all this is making you sick, just pretend this frog is actually the guy who invented car alarms, but he’s been put under a terrible thousand-year spell by a sleep-deprived witch.)

The first equalizer most of us ever played with only had two bands (regions of the spectrum that you can adjust); many home or car stereos, in lieu of something actually labeled “eq”, have a pair of knobs labeled bass and treble — which is still an equalizer. There’s generally a large gap in the spectrum between these two bands, so you can effectively adjust the midrange by turning both bass and treble up and down together, and then re-adjusting the overall volume. Hopefully, you already have a feel for how to use these. If not, play with them. And not conservatively — set them at extreme positions, and listen for a while.

Done with that? Good, now take off your training wheels and hop onto the ten-gear, er, I mean, ten-band. Let’s look at a typical set of bands again (the exact frequencies on yours may deviate from this):

30 - 60 - 125 - 250 - 500 - 1000 - 2000 - 4000 - 8000 - 16000

Each number tells you the frequency in the center of that band, but that slider will affect frequencies below and above it to progressively lesser degrees, somewhat overlapping the range of the next band over. On a ten-band equalizer, the spectrum is divided up into about one octave per band. That’s because our hearing range is approximately ten octaves. As I said in the last installment, each octave is double the pitch of the octave below it, so as you go from band to band, the frequency approximately (if not exactly) doubles. Some equalizers of this type split the spectrum into more bands: 15, 20, or even 30. While easy to understand visually, these are more time-consuming to adjust. So other types of equalizers and filters have been invented, such as parametric equalizers and FFT filters, to give you more precise control with fewer parameters — but since those require more experience to use effectively, we’ll stick to an ordinary ten-band for now.

Most musical sounds will be spread over several of these bands. This is partly due to the range of the instrument, but also due to its harmonic content. A single note on an instrument, played all by itself, doesn’t just contain energy at one frequency — that note contains many harmonics, also called overtones, which are higher than the note itself (the fundamental), and give the sound its color, character, or timbre. In most cases, these overtones are exact whole-number multiples of the fundamental, and they don’t sound like extra notes, because you just hear them as part of that note’s sound.

If you’re a guitarist, you might know that you can isolate harmonics by touching the string at certain points while you play it. It’s important to realize that those higher pitches are actually in the note regardless; if you listen carefully while alternating between playing the harmonic and playing the open string normally, you can hear that those higher tones were in there all along. All you’re really doing is muting some of the harmonics, including the fundamental, so that other harmonics are now more prominent.

So if you have a recording with an unwanted note in it, at about 250 Hz — this would be close to the open B string on a guitar — you could try to remove that note by cutting the 250 Hz band on your equalizer, but that would still leave its overtones at 500 Hz, 750 Hz, 1000 Hz, 1250 Hz, and so on. Just from hearing all those harmonics, your brain will fill in the phantom fundamental and you’ll still perceive it as the same note, albeit more trebly. This is not to say you could never use filtering to soften a ringing open string on a sloppily-played guitar track, but its use for that type of thing is limited and difficult. (I’ve done this in extreme circumstances, by making four or five very narrow cuts on an FFT filter, right where the first few harmonics are, since they’re generally the loudest.)

The best way to think of equalization, though, is as an overall sound-shaper. When you split the sound into those ten (or however many) bands, you can think of them as “parts of the sound”, but you have to think of those “parts” in a different way than as specific instruments. There are no “good parts” and “bad parts” — though the average consumer is more likely to buy a stereo (or a CD) if they hear proportionately more extreme bass (below 50 Hz) and extreme treble (above 8 KHz), the meat of your music — melody and harmony (remember those?) — is always somewhere in the middle, and you do more or less want to give your listener a balanced meal, right?

Get to know the spectrum. I’ll introduce you to a few slices of it, but then you need to go and play with an equalizer and form your own mental relationship with it. All the numbers below are “general areas”, and there is no hard-fast cutoff point where one really begins and another ends, so don’t try to memorize so much as get the general idea.

Like I said, below 50 Hz is super-low bass, the kind that some drivers like to generously share with us while they still have their hearing. The 100 Hz area is still bassy, but not so bassy that most males couldn’t sing it. If you have a track with a higher pitched instrument, like a flute, and that same track has some unwanted rumbling on it, you can cut these lowest frequencies to get rid of the rumbling without hurting the instrument.

100-300 is what I might call upper bass or low midrange, important because a lot of fundamentals are happening here, as well as the “oomph” of the snare drum (its lowest tone, the part of it that you can almost “feel”, although snare drums have energy across pretty much the whole spectrum). Too much of this will sound muddy, but too little will sound hollow.

Going from there to about 1000 Hz will take you to the tippy-top of the female singing range, and many of the highest notes on ordinary instruments such as a guitar.

Most of what you hear above 1000 Hz are overtones from lower notes, which are very important to the character of those notes. If you listen to speech through a band centered at about 2000 Hz, it sounds like it’s through a bullhorn. Through a 4000 Hz band, it sounds more like a telephone. These are important “parts” of the voice, though, and if you removed them, the voice would sound muffled. These upper-midrange areas are what we normally think of when we say “tinny” or “mid-rangey”, and the “bite” at the top of an electric guitar is generally in here.

When you get to about 8000 Hz and up, this is where the cymbals mostly are, the sibilance (”ssss”) and breath on vocals, and the clarity of acoustic guitars. Some darker or more mid-rangey sounds, like bass guitars and electric guitars, don’t have much going on in this range, so it’s easy to remove annoying hiss from those tracks just by cutting the highest frequencies. Also if you took the sound directly out of a distortion pedal without using an amp, you might cut these highs just to make it warmer and less fuzzy. You can sometimes liven up a dull sound by increasing this, but watch out for hiss.

In the actual practice of recording your own music, you will first want to try to get your sounds right as you go. But we all make mistakes, and in those cases you can sometimes improve a track by equalizing it. Equalization is also used on entire mixes, subtly and carefully, as part of the mastering process.

Your homework is to go on a date with an equalizer, with hopefully ten bands, or close to it. It can be an old-skool physical beast, or it can be the one that comes with iTunes, Winamp, or whatever. You’re not going to use it on this outing, you’re just going to get to know it. Bring your favorite romantic music — preferably something with drums and electric bass, and then maybe some acoustic songs on it. Turn all the bands down low, and then one at a time, turn a band all the way up by itself (if it’s too loud, turn the overall volume down) and just listen to it. What’s in there? What part of the sound is in that part of the spectrum? What does it mean to you aesthetically, and how does it make up part of the big picture? Leave that band on long enough to hear how much of the drums come through, and how much of the other instruments come through. Then turn it down and go on to the next band, and do the same. Don’t be in a hurry… just listen, and sort of meditate on the “slice of sound” that you’re listening to. Don’t play favorites or give preferential treatment; remember, they all go together to make up the music you know and love.

Next, set all the bands to their normal position, and this time, cut one band at a time. If you’re wearing headphones or have small speakers, you might have trouble hearing the difference when you cut the very lowest one — if so, don’t worry about it too much. Then bring it back to normal and cut the next one. What’s missing from the overall sound now? Do this for all the bands. Get a feel for what it’s like when energy is missing in each part of the spectrum. As you do this, there may well be times when you find yourself actually liking some of the cuts better than the original sound. Please don’t say you’re going to cut that band out of every song you ever do for the rest of your life! You may have a tendency to get emotional, but you and the equalizer are just friends.

When the date is over, thank the equalizer, and in a polite, rational way, say “I plan to use you from time to time, but I’m going to be using other tools as well. It’s nothing personal.” The equalizer will actually respect you more for this, and your recordings will reflect the sophistication of this relationship.

The song that you sliced up like a salami may not be quite as happy. Hey, what can I say, that’s science for you.

*Ribbit*