Basics of Synthesis

Let’s learn the parts of a synthesizer! Whether you have Serum, a Juno, Massive, an MS20, FM8, a Minimoog, Helm, an Easel, Phase Plant, or really any synth, it’s good to know what you’re dealing with. Most synths, you’ll find, have the same parts:

Oscillators

Oscillators can be as basic as having the 4 basic shapes (saw, triangle, sine, pulse) and 3 basic modulations (FM, sync, PWM). Most oscillators like these are found in the analog world, but are still amazingly powerful when you start making use of their idiosyncrasies. They’re generally either saw- or triangle-core, meaning the basic waveform the oscillator creates is either a saw wave, derived from an integrated pulse train (basically a series of pulses that smooths the transition from on and off), or a triangle wave, derived from an integrated square wave. Triangle-core oscillators are often more stable than saw-core, but saw-core can be more capable of weirdness, which is fun for sound designers.

Oscillators are composed of waveforms, which have 3 primary aspects: frequency, amplitude, and phase; or, how often a wave cycles, how much it cycles, and when it cycles. Frequency is how you set musical pitches, but can also be modulated at audio rate to create more harmonics, at the expense of losing any semblance of tuning. Amplitude is how loud the signal is; or, further, how loud the signal is over time. A saw wave’s signal goes down over time following each pulse, whereas a sine wave is continuously going up or down. Phase can be reset with sync, but sync isn’t phase modulation; rather, phase modulation requires the continuous changing of phase over time, rather than a hard reset of phase per cycle.

Each core waveform, saw or square, can be further waveshaped. Using a saw core as an example, you can put a comparator on that saw wave to create a square wave. modulating the comparator’s threshold will vary the pulse width of the square wave – PWM, if you will. You can further integrate that square wave to create a triangle wave. You can then clip and filter a triangle to make a sine, albeit not a perfect one!

So far, I’ve only discussed analog oscillators and how they work. But, so many other oscillators exist, like wavetables, which take an array of waveforms and smoothly transitions between them. I’ll talk more about these later.

Filters 

Filters are used primarily in subtractive synthesis, but have cropped up in other synthesizers as well. They remove, or filter out, parts of the sound. The 4 basic filter types are lowpass, which only passes low frequencies/filters out high frequencies, highpass, which does the inverse, bandpass, which only passes a band of frequencies, and notch, which does the inverse.

Filters typically only have 2 controls: cutoff and resonance. Cutoff determines the frequency at which the filter starts to work; for example, if a lowpass filter is set to 10khz, anything above that will be removed, with the rest being largely unaffected. Resonance creates a resonating peak at the cutoff frequency, emphasizing that frequency. This works by feeding the filter output back to the input; because of this, high enough feedback levels can create oscillation!

Filters also have different slopes they can operate at. Most synthesizers use 12db slopes, meaning the filter removes signals by 12db per octave. FIlters don’t immediately cut signals due to this, though you can get close with very high slope filters, like 96db slopes. Furthermore, filters cause a 90° phase shift of the entire signal. Sometimes this is desired, such as with phasers, which use allpass filters (pass all frequencies, but change phase). You can also combine filter types and slopes to create new filter types, as was used extensively with the Oberheim Matrix synthesizers.

Multiple filters together can create unique sounds, such as formants with separated cutoff frequencies with high resonance; comb filters, which use an allpass mixed with the original signal (or a series of bandpass filters); and equalisers, which use a series of limited lowpass, bandpass, and highpass filters with vary levels to alter and sculpt sound.

Amplifiers

Amplifiers take the filter’s result and change the volume. they often don’t actually amplify anything, and rather attenuate the signal based on a control signal, often an envelope generator. VCAs, the devices which do this in analog synths, or voltage controlled amplifiers, can be used for more than just final sound sculpting as well. My mixing a saw LFO with a decaying envelope, you can create a delay effect. By modulating amplitude at audio rates, you create amplitude modulation synthesis, sometimes called granular formant synthesis, or packet synthesis. Polarisers, VCAs that invert the signal as well as attenuate it, can be used for ring modulation as well as frequency shifting, plus they can modulate control signals!

Envelopes

Envelopes are one of the more common modulation signal generators, and generally have 4 stages: attack, decay, sustain, and release, or ADSR. This setup is so common, many people erroneously call all envelope generators ADSRs. However, envelope generators with more, and less, stages, are common, including AD, AR, ADDSR, and… well, you can basically add or remove however many stages you want. You can also often control each stage’s time with a control signal, and can sometimes add a delay to the envelope, so it starts generating after a set time.

Controlling volume and filter cutoff with envelopes is most common. Controlling oscillator pitch is also useful for creating drum sounds and dynamic sync sounds. In modular environments, you can often use envelope generators as waveshapers by plugging an oscillator into it and shaping the resulting waveform with the envelope controls.

LFOs

LFOs, or low frequency oscillators, are another common modulation source. They generally act exactly like normal audio oscillators, but at sub-audio rates, allowing for slower modulations over time. Things like old-school dubstep wobbles can be created by modulating a lowpass filter with an LFO. Tremolo effects are created by modulating a VCA with an LFO. Vibrato is done by controlling oscillator pitch with an LFO. YOu can also create more complex LFOs by modulating one’s frequency with another’s.

If all synths have the same parts, why use more than one? All synths sound different and do things slightly differently. For example, Serum uses samples of waveforms and loops them while morphing between waveshapes and filters them, while a Minimoog uses negative feedback analog circuitry to create hysteresis at audio rates, causing oscillation which is then sent into a saturating mixer and ladder lowpass. Same parts done differently creates totally different results.