Jim Aikin’s Modular Synthesis for Beginners – Filters: When Less Is More

Modular Synthesis for Beginners…

Modular Synthesis for Beginners with Jim Aikin

When synthesizers first started to take off, back in the ’60s and ’70s, oscillators were capable of producing only a few basic waveforms…

On their own, the sawtooth and square/pulse waves sounded harsh—good for industrial and glitch music (though those genres hadn’t yet been invented) but not much else. To tame these waves into a more musically useful form, a synth needed a filter.

A filter lowers the level of some of the frequency components of a tone while letting others pass through. Because it removes part of the incoming tone, using a filter is referred to as subtractive synthesis. These days, other ways to make electronic sounds have become more common and more versatile, so subtractive synthesis with filters is maybe slightly old school, but it’s still incredibly useful.

Because a filter removes (or reduces the level of) some part of the signal that’s being filtered, it can only do its job if we feed it a signal that has some energy at a variety of different frequencies. If you send a pure sine wave through a filter, the filter will have almost nothing to chew on.

Explaining in words what a filter does can mean slugging it out with some heavyweight terminology. It’s a lot easier to hear what a filter does than to explain it. If you’re new to synthesis, you’ll probably learn more by watching this month’s video than by reading the text. But let’s define some terms.

As usual, I’ll be using VCV Rack in the video. If you haven’t downloaded and installed it yet, you can find instructions HERE. This month I’ve set up two patches, one for basic demo purposes <Filters 1.vcv> and one to show how to make a notch filter <Filters 2.vcv> if you don’t have a notch filter.

A Glossary of Filter Terminology

Frequency Components. Most sounds have some energy at various frequencies. The places in the frequency spectrum where the sound has energy are its frequency components. In the case of a waveform such as a sawtooth wave, we call them partials. A nonperiodic sound is called noise. Noise signals have energy at all frequencies, so we can’t say that noise has partials. A sine wave has only one partial (its fundamental frequency).

Lowpass Filter. A lowpass filter allows the lower frequency components of the incoming signal to pass through, while filtering out the components that are at higher frequencies. If there is no sound energy at the higher frequencies, the filter will do essentially nothing.

Highpass Filter. This is the opposite of a lowpass filter. A highpass filter allows the higher frequency components of the signal to pass through, while blocking the components that are lower in frequency.

Cutoff Frequency. This parameter sets the frequency at which the filter will start to have an effect on the signal. Sometimes it’s called just plain cutoff or just frequency. The real mathematical definition is a bit more complex; technically, the cutoff is the frequency at which the filter is attenuating the signal by 3dB. But that fact will be of interest only if you’re designing a filter, not if you’re using one to make music.

Resonance. Most analog-type filters have a parameter called resonance; older terms are emphasis and q. The resonance parameter introduces a peak in the frequency response at the cutoff frequency. Components of the signal that are at the cutoff frequency are actually boosted to a louder level. When the resonance is set high enough, a filter may self-oscillate, producing a tone even when there’s no input signal.

Attenuation. Attenuation is a fancy term for lowering the amplitude—and, of course, amplitude is a fancy term for the loudness, strength, or level of a signal. When a signal is attenuated, its level is reduced. A standard synthesizer filter doesn’t actually block frequency components that are beyond the cutoff frequency. It attenuates them.

Rolloff Slope. This is easiest to explain by limiting the discussion to a lowpass filter. The further above the cutoff frequency a sound component (a partial) is, the more it will be attenuated. The amount of attenuation is a function of the rolloff slope. Typical rolloff slopes are 6, 12, and 24dB per octave. What this means is that if the lowpass filter has a 6dB per octave rolloff slope, a partial that is one octave above the cutoff frequency will be attenuated (lowered in level) by 6 decibels. If it’s two octaves above the cutoff, it will be attenuated by 12 decibels, and so on. This is a shallow rolloff slope. If the lowpass filter has a rolloff of 24dB per octave, the effect of the filter will be much easier to hear. A partial one octave above the cutoff will be reduced by 24dB, a partial two octaves above the cutoff by 48dB, and so on.

Another way of talking about the rolloff slope is to refer to the filter’s poles. A two-pole filter (sometimes abbreviated “2P” in a drop-down menu) has a rolloff slope of 12dB per octave, while a four-pole filter has a rolloff of 24dB per octave.

Bandpass and Notch Filters. A bandpass filter allows a band of frequencies somewhere in the midrange to pass through, while attenuating both the higher and lower frequencies. A notch filter, also called a band-reject filter, is just the opposite: it reduces the level of frequency components within some band (called the notch) and lets both the higher and lower frequencies pass through.

If you don’t have a bandpass or notch filter module in your system, you can produce the same result with a highpass and a lowpass filter. To produce a bandpass, chain the lowpass and highpass filters in series. To produce a notch effect, configure them in parallel, as demonstrated in the video you see above.

Filter Modulation. When using a filter in a musical context, you’ll quite often want to modulate the cutoff frequency with an envelope. The filter will most likely have a dedicated modulation input for this type of effect.

In preconfigured (non-modular) synths, you’ll usually find a second dedicated input to allow the cutoff frequency to track the keyboard. When you play higher on the keyboard, the cutoff will also rise. This will tend to produce a more uniform tone across the keyboard. If you’re using VCV Rack, however, you’ll find that most of the VCV filter modules have only a single input for modulating the cutoff. In order to get both envelope and keyboard (or step sequencer output, or LFO) modulation, you’ll have to use a mixer to combine the two modulation signals.

Drive. Many filters these days have a drive parameter. When you add drive, the level of the incoming signal is boosted. Drive will add a bit of subtle distortion to the signal, producing a richer sound. It may also boost the output level, so when you add drive you may need to reduce the level of the signal in your mixer.

More Specialized Filter Types. Some filter modules offer a variety of types, or modes of operation, such as comb filtering. A comb filter introduces a bunch of resonant peaks, which can be shifted up or down to add movement to the sound. Less common, but sometimes useful, is the formant filter. A formant filter is configured with several bandpass filters in parallel. The filter will produce peaks at several frequencies, thus mimicking the acoustic behavior of the human vocal tract. A formant filter may have settings for common vowel sounds, such as ah, oh, oo, and ee.

A state-variable filter can morph smoothly from lowpass through bandpass to highpass operation. In some commercial synths, you’ll find even more complex types—lowpass/notch hybrids, for example. Using filters opens up whole worlds of sound.

View More Synth Academy