At the most basic level, a musician controls three elements: pitch, time, and timbre. An engineer's job is to help (or sometimes second-guess) musicians in this quest. Although minute control over pitch and time has been within our grasp only in recent years, controlling timbre is a well-established part of engineering. Each time you choose a microphone or tweak a preamp, you're affecting the recorded timbre. The tool we most often think of for timbral processing, however, is the equalizer.
FIG. 1: A graphic equalizer divides the audio spectrum into 30 or 31 bands and provides clear, concise visual feedback of what has been cut or boosted.
An equalizer, or EQ, is a hardware or software signal processor that allows you to manipulate a sound's frequency spectrum. If you've ever used the tone controls on your home receiver or made a smiley face out of the sliders on the front of your car stereo, you know what EQ does — just enough to get you in trouble. In this column, we'll look at the various components of an equalizer, explore how they work, and describe how they can be applied to improve your sound.
EQ Pluribus Unum
To paraphrase the 18th-century French mathematician Joseph Fourier, any complex sound is made up of a number of different frequency components. Because this is true, we can change a sound's basic timbre by manipulating those components independently. That is the function of an equalizer: to amplify or decrease portions of the frequency spectrum. This is accomplished with filters (for a discussion of filter basics, see “Square One: Just Passing Through” in the February 2005 issue of EM, available online at www.emusician.com). As you may know, a filter is also an essential part of a synthesizer. Although a synthesizer filter is designed to impart a unique character to the sound, and an EQ filter is more often designed to do its job inconspicuously, they are in fact the same thing.
Like synthesizer filters, EQ filters come in various types. An equalizer can combine several regions or bands of filters within a single processor, and most do. That smiley face on your car stereo is really seven or more bandpass filters with fixed bandwidths and center frequencies. Because the sliders give obvious visual feedback, this is known as a graphic equalizer (see Fig. 1). Professional graphic EQs offer 30 or 31 bands, each affecting a range of one-third of an octave. Such devices are standard equipment in sound reinforcement, as they give the engineer easy and direct control over any frequencies that start to feed back during performance. A skilled front-of-house engineer can listen to a room and visualize the graphic EQ's curve before touching a control.
FIG. 2: A multiband equalizer''s lowest and highest bands often default to high- and lowpass filters, respectively.
Although graphic EQs are also found in studios, a more common design includes four to seven bands of different types. Software EQs often allow the user to disable unused bands to conserve resources or to change filter types for each band. Although each band is technically a filter, when an engineer refers simply to “the filters,” she's probably talking about the first and last bands, which are often fixed or defaulted as high- and lowpass filters, respectively (see Fig. 2). In addition to variable cutoff frequency, filters normally offer variable slope, which refers to the steepness of the filter. The slope could be continuously variable, or it might be adjustable in increments of 6 dB up to 24 or even 32 dB per octave. In the latter case, the term pole is typically used for each 6 dB increment — a 24 dB-per-octave slope would be a 4-pole filter.
Filters are most often used simply to restrict the bandwidth of a track. After all, if you're recording piccolo, anything below 500 Hz is by definition not part of the piccolo's sound. Filtering the low end reduces bleed from other instruments, helps mitigate any mechanical noises around the mic, and helps prevent environmental sounds such as air-conditioning and electrical hum from accumulating on multiple tracks. Filtering the high end of a kick drum reduces bleed from the snare and cymbals, keeping the kick track as clean and isolated as possible (see Web Clip 1).
Engaging both high- and lowpass filters creates a bandpass filter. Although on some EQs the filters have limited range, many EQs allow the high- and lowpass filters to cover most if not all of the audio range. Such an EQ would enable you to place the two filters very close to each other, creating the standard “telephone” filter that's used on so many lead vocals and drum loops (see Web Clip 2).
Often the top and bottom bands of an equalizer can be switched from high- and lowpass mode to a shelving mode. A shelving filter boosts everything above (or below) the cutoff frequency by an equal amount — its effect does not increase with frequency (see Fig. 3). The treble and bass controls on your home stereo are most likely shelving filters with a fixed corner frequency. On an EQ, shelving filters ordinarily offer a variable corner frequency, but their use is essentially the same — to boost or cut “the bass” or “the treble” of a track or mix. For example, a high shelving filter would be just the thing to add sizzle to a synth pad or string ensemble (see Web Clip 3). Shelving filters may offer control over only the corner frequency and amount of boost or attenuation, or they may also allow the user to adjust the slope of the transition band.
FIG. 3: A shelving filter''s effect is not progressive like that of a lowpass or highpass filter. Instead, it boosts all frequencies above (or below) its corner frequency equally.
The remaining bands of equalization are ordinarily made up of bandpass/band-reject filters, which do exactly as their names suggest, affecting a range of frequencies. The affected range is defined by a center frequency and bandwidth. Bandpass filters are used for a great variety of purposes, and some sounds may require the use of several bands. Unlike those found in a synthesizer, however, an EQ's bandpass filters do not block all frequencies outside the band, but rather provide a boost or cut within the band without affecting other frequencies. The alternative name peak/notch filter is arguably more precise.
The most flexible bandpass filters are called parametric, meaning simply that each of their parameters (frequency, bandwidth, and gain) is independently adjustable. A band whose controls are not fully independent — for instance, one whose bandwidth is linked to its center frequency — may be called quasi- or semiparametric.
In principle, the use of a bandpass filter is quite simple: boost what you like, cut what you don't. In practice, this takes a skilled ear and refined judgment. A good way to develop your ear is to “sweep” the EQ. Boost a band as much as you can, and narrow its bandwidth as far as you can. If the EQ provides a graphic display, you will have created a very sharp spike. Now slowly raise and lower the band's frequency, listening carefully. At some point, you will hear either an attractive or an unpleasant portion of the track jump out at you. Note that in a host-based DAW, latency may cause you to overshoot the target frequency, so take your time zeroing in.
Once you've found a key frequency band, adjust the amount of boost or cut and the bandwidth to a more appropriate level — no matter how attractive the frequency, you probably don't want the band maxed out. Repeat this for as many bands as needed.
It's common practice to rely on cutting bands when possible, for reasons of gain staging and filter character. When boosting a band, make sure the increased gain does not distort the subsequent circuit. If the EQ provides output meters, pay attention to them. Also listen for the edgy sound of phase distortion, and consider whether a complementary cut in a similar sound would serve your purpose as well. For example, when a kick drum and bass part are getting in each other's way, a wise engineer looks first to trim away part of each to let the other shine through.
The term Q is sometimes substituted for bandwidth or slope. You may remember this as the term for resonance in a synthesizer filter, and in an EQ filter, it means essentially the same thing. However, EQ filters are more typically (though not always) crafted without resonant peaks even at very steep slopes and narrow bandwidths. Remember that a high Q value counterintuitively means a narrow bandwidth or steep slope.
Whatever EQs you have at your disposal, whether software or hardware, no matter the design, these essential principles hold true. Get to know the sound of different band types, slopes, and bandwidths, and tune your ear to pick out the good and bad frequency components. Listen, experiment, and listen some more, and you'll be a master of timbre in no time.
Brian Smithers is course director of audio workstations at Full Sail Real World Education in Winter Park, Florida.