Synthesis Techniques

Synthesis is a vast and varied topic. I’ll explain what techniques exist and how they function in this article.

Subtractive is the most common, and described in detail in my “What’s a synthesizer” post. It involves taking a sound, generally from an oscillator, and subtracting parts of that sound, generally via filtering.

Additive is an interesting method of synthesis that involves taking a bunch of sine waves arranged to harmonic intervals and varying their volume to create new sounds. Additive is cool because it can create literally any sound you can possibly imagine, given that every sound has a fundamental and harmonics of varying levels above it! It sucks because it’s a pain to use, since each harmonic is individually controlled over time. Modern additive synthesizers tend to try to simplify this, often removing the possibility of many unique aspects of additive, and thus its ability to create any sound, limiting you to the sonic palette of the creator.

Wavetables are extremely common now, despite being technology from the 70s. The idea is each cycle of a waveform can be a sample, a short audio recording, looped at audio rate. Originally, each waveform was chosen individually. In the 1990s, interpolation (crossfading) between waveforms was created to allow for morphing between waveforms. This eventually gave rise to synthesizers like Serum and Massive, which are used heavily in modern genres of electronic music. Further developments of wavetables within the modular community enabled multi-dimensional wavetables, allowing for extremely complex sounds to be created with a single oscillator. Usually wavetables are implemented as oscillators within an east-coast subtractive synth.

Waveguides were invented in 1983 in the form of Karplus-Strong synthesis. It’s a digital synthesis method now mostly used for physical modelling, but can of course be used for otherworldly and utterly new sounds. The basic principle is thus: a burst of noise feeds a delay effect with delay time set to audio rate. A lowpass filter is inserted into the feedback loop to dampen the sound over time. This creates a plucked string sound, called Karplus-Strong string synthesis. Building upon that, we can create what is called modal synthesis, a physical modelling synthesis method that involves a synthesized exciter feeding a delay effect with delay time set to audio rate. An array of bandpass filters with adjustable cutoff and volume is inserted into the feedback loop, creating unique dampening or excitation effects as time progresses. I use this for the majority of my sound design when it comes to weird and realistic sounds.

Waveshaping is core to the west-coast synthesis method, as discussed briefly, but has uses outside of that. Most people think of waveshaping and think of wave folding, which is the most common form of interactive waveshaping and quite literally folds a waveform atop itself. However, even the 4 basic waveforms are derived from waveshapers. Distortions and filters are types of waveshapers as well. My personal favourite waveshaper is an analog eurorack module from Doepfer, the A-136 distortion/waveshaper, because it allows for different kinds of ‘clipping’ like distortion, ‘folding’ like wavefolders, ‘shaping’ like triangle to sine shapers, and everything in between. You can also use waveform manipulation, such as that seen in Mutable Instrument’s Tides and Tides mk2.

FM is a lovely topic, often of much debate due to mis-use of the term in the 80s which has continued to this day from Yamaha and their DX, and later certain SY and TG, synths. Frequency modulation and phase modulation sound very similar when using sine waves and all else is equal, with the difference being with FM, base frequency changes, and PM retains base frequency. This means you can use PM in place of FM for a similar sound that stays in tune. Problems began to arise when waves other than sines were used, at which point the sound alters significantly from phase modulation, which is why PM is used so extensively even in place of FM, such as in Serum. FM8 is another popular synth that uses ‘FM synthesis’, or phase modulation. 

Granular is one of my most-used synthesis techniques for pseudo-organic sounds, like monster noises. It involves taking audio and chopping it up into tiny granules (hence the name). Those granules can be of varying length, shape, pitch, and timing, which allows something like a simple saw wave to turn into a monsterous growl, or polyphonic pad, or even a multi-part drumkit.

Linear Arithmetic is a fancy word Roland came up with to describe the combination of a sample and synthesized sound, generally used to create more complex sounds. Originally it was used to create more realistic sounds, by combining a transient, or attack, with a synthetic decay, or body, such as that of a hit drum or plucked string. It turns out you can make way crazier sounds by just mixing a sample and a synth, which is most fully realised in synths like the SY77 and Serum, rather than Roland’s own D-50 and similar.