Some synths include a resonant lowpass filter among their effects algorithms. Because all of the synth voices are summed to a mono or stereo signal before being passed through this filter, and because the synth voices may start and end their notes at different times, it doesn’t make much sense to modulate filter cutoff from any of the voice envelopes. Instead, the filter effect usually incorporates a device called an envelope follower. An envelope follower senses the amplitude of an audio signal (in this case, the signal entering the effects processor) and outputs a control signal with a corresponding amplitude. The control signal, in turn, modulates filter cutoff. As a result, when you play the keyboard harder or play more notes, the envelope follower will open the filter cutoff further. As the sound coming from the voice section dies away, the filter cutoff will fall. In many effects processors, this is called an auto-wah effect.
Like other lowpass filters, a filter effect will probably have parameters for cutoff frequency, resonance, and envelope amount. The envelope follower will probably have two parameters: attack and decay or release. In this case, “decay” and “release” mean essentially the same thing. The attack parameter controls how quickly the envelope can rise as the amplitude of the incoming signal increases:
With a fast attack time, the envelope will respond more quickly. The decay parameter, in turn, controls how quickly the envelope can fall back toward zero when the amplitude of the signal decreases.
You might expect that it would be a good idea to set both attack and decay to short time values (fast attack, fast decay). But if you do this, the output of the envelope follower will tend to be a little jumpy.
Most often, a fast attack should be combined with a medium-to-slow decay. In some instruments, these are the defaults, and you can’t change them.
Equalization
Equalization (better known as EQ) has been used in recording studios for decades. An equalizer can boost or cut the frequencies in selected parts of the frequency spectrum while leaving other parts of the spectrum untouched. Unlike many of the effects discussed in this chapter, which can alter sounds in spectacular ways, an equalizer is more or less intended to improve the sound without being noticed.
Equalization is used most often not for enhancing or controlling the timbre of individual sounds by themselves, but for shaping their frequency content so they’ll “sit better” in a mix with other sounds, either onstage or in a recording situation. Because of this, some synths put their EQ not in the effects section, where it would be programmable separately for each patch, but in the global output section, so that the EQ settings won’t change when you select new patches.
With an equalizer, you’ll be able to define the frequency range to be affected and the amount of cut or boost to be applied to partials that fall within that range. The amount of cut or boost is usually specified in dB — a range from -18dB to +18dB may be about average. Other than that, there’s not much to say about cut and boost. There’s more variety in the way the frequency range is defined.
For each range of frequencies to be affected, we can talk about the bandwidth and the center frequency (see Figure 9-1). The bandwidth is measured in Hertz. For instance, if we want to boost frequencies between 800Hz and 1,200Hz, the bandwidth would be 400Hz. The center frequency of this band is 1,000Hz (1kHz).
A graphic EQ (see Figure 9-2) provides a number of frequency bands (10, 15, or more) of boost/cut.
The bandwidth and center frequency of each band is fixed, in such a way that all of the bands together cover the entire audible frequency range. Each band might cover one octave, for instance. In that case, the center frequencies might be 24Hz, 48Hz, 96Hz, 192Hz, 384Hz, and so on. If there are more bands, each band can have a narrower bandwidth, thus giving you more control over the frequency spectrum.
Figure 9-1. The frequency response of one band of a parametric equalizer is governed by three controls.
The cut/boost parameter changes the height of the “bump” in the response curve, the center frequency parameter moves the bump from left to right, and the bandwidth parameter changes the width of the bump.
Figure 9-2. A hardware graphic equalizer. The term “graphic” comes from the fact that the front-panel sliders provide a visual representation of the frequency response.
Graphic EQs are seldom found in synthesizers. More often, you’ll see a parametric or semi-parametric design (or some combination of semi-parametric and semi-semi-parametric). A semi-parametric EQ provides fewer bands, but gives you more control over each band, the assumption being that for the most part you’ll want to leave the signal untouched. A single parametric band may be all you need to dial in the required boost or cut for a troublesome sound.
A parametric EQ band contains three controls: boost/cut amount, center frequency, and bandwidth. A three-band parametric EQ (not an uncommon design) provides three independent bands, each with these three controls.
A semi-parametric EQ band has only two controls: boost/cut amount and center frequency. The bandwidth is not directly controllable. Some synths, for example, provide EQ effects with two fully parametric bands for processing midrange frequencies, plus semi-parametric high and low shelving bands. A shelving band (so called because the EQ curve looks rather like a shelf, as Figure 9-3 indicates) can cut or boost the lows or highs, but it will cut or boost all of the partials that fall above or below the frequency set by the frequency parameter. This parameter is sometimes called the corner frequency, because the frequency response curve has an angled shape. With a low shelf EQ whose corner frequency is set to 150Hz, for instance, all of the partials below 150Hz will be cut or boosted. In effect, the frequency parameter of a shelving EQ band controls both the bandwidth (which in this case goes from 0Hz to 150Hz) and the center frequency (75Hz), which is halfway between the corner frequency and the outer end of the frequency range.
Many EQ algorithms in synthesizers limit the parameter settings in various ways. For instance, you might find two fully parametric mid-bands, one whose center frequency can be set between 100Hz and 1,000Hz and the other with center frequency settings between 800Hz and 8kHz. The low and high shelving might have no frequency parameters at all, only cut/boost parameters.
When using EQ, it’s important to understand that an equalizer can neither cut nor boost overtones that don’t exist in a given signal. If a sound has no high-frequency content to begin with, cranking up a high shelving EQ will only add a bit of background noise. To use EQ effectively, you need to develop a sensitivity to what’s going on in various parts of the frequency spectrum.
Removing DC Offset. In some programs, such as Csound and Max, that will do experimental digital synthesis, certain types of DSP, such as digital reverbs, may introduce a DC offset into the signal. A DC (direct current) offset is a nasty form of distortion: While inaudible for the most part, it causes clicks and pops at the ends of sounds. DC offset can be spotted in a waveform display. As a sound decays toward silence, the waveform won’t converge on the center line but will instead taper to a line above or below the center line.
Figure 9-3. A low shelving band in an equalizer (left) affects all of the frequencies below the corner frequency. A high shelving band (right) affects all of the frequencies above its corner frequency.
Since DC is by definition a partial with a frequency of 0Hz, a low shelving EQ or highpass filter set to an extremely low frequency (such as 5Hz) will remove DC offset without affecting the sound of the music.