Types of Synthesis Explained: Subtractive, FM, Wavetable and More
Every synth sound you've ever heard, from a Minimoog bass to a Serum growl, comes from one of a handful of ways of making sound electronically. Understand those and you'll program presets faster, pick the right synth for the job and stop wondering why a pad patch on one synth won't translate to another.
Here are the main types of synthesis, how each one works, what it sounds like and which synths use it.
A very short history
Electronic instruments go back further than most people think. The Telharmonium was making music from giant dynamos around the turn of the 20th century, and RCA's room-sized Mark II synthesizer was running at the Columbia-Princeton studio by the late 1950s. The modern synth really arrived in the mid-1960s, when Robert Moog and Don Buchla independently built voltage-controlled modular systems, and the Minimoog of 1970 put that sound into a keyboard you could carry. Almost everything since has built on, or reacted against, those designs.
Subtractive synthesis
Subtractive synthesis starts with a harmonically rich waveform, usually a sawtooth, square or pulse, and carves it with a filter. A low-pass filter takes out the top end, and sweeping its cutoff is the classic "wow" of analogue synths. An amplifier shaped by an envelope then decides how the sound starts, holds and fades, while LFOs add movement.
It's the most intuitive kind of synthesis to learn, because every control does something you can hear straight away. It's brilliant for basses, leads, brass and pads, and it's the sound of the Minimoog, Prophet-5, Juno-106 and most of the analogue synths that defined electronic music.
In software: u-he Diva, u-he Repro, TAL-U-NO-LX and Cherry Audio DCO-106.
Additive synthesis
Additive synthesis works the other way round. Any sound can be described as a set of sine waves at different frequencies and levels, its harmonics or partials. Additive synths build sounds by stacking those sine waves and controlling each one, often with its own envelope.
In theory it can make anything. In practice, setting dozens or hundreds of partials by hand is hard work, which is why most additive synths offer macro controls or resynthesis, where the synth analyses a sample and rebuilds it from sine waves. The sound tends towards glassy, organ-like tones and spectra that shift and shimmer over time.
Examples: the Kawai K5000 hardware synth, Image-Line Harmor and Native Instruments Razor.
FM synthesis
Frequency modulation synthesis was developed by John Chowning at Stanford in the late 1960s and early 70s, licensed to Yamaha, and launched into the mainstream by the DX7 in 1983. It uses one oscillator, the modulator, to modulate the frequency of another, the carrier. When the modulator runs at audio rate, it creates new sidebands, so two plain sine waves can produce a bright, complex tone.
FM synths call each oscillator an operator and arrange them in algorithms, which set who modulates whom. The ratio between operators decides whether the result is harmonic, like an electric piano, or clangorous, like a bell. FM is famous for being tricky to program, but it's unbeatable for electric pianos, bells, plucks, metallic percussion and hard, punchy basses.
A close cousin is phase distortion, used in Casio's CZ synths, which bends the shape of a waveform's cycle to similar effect.
In software: SONiCBITS Exakt Lite (free), GForce Halogen FM and u-he Bazille, which combines FM and phase distortion in a modular layout.
Wavetable synthesis
A wavetable is a series of single-cycle waveforms lined up in order. A wavetable synth plays one frame at a time and lets you move through the table, so modulating the position morphs the sound from one waveform to the next. Wolfgang Palm pioneered the idea, first in his PPG Wave synths of the early 1980s and later with Waldorf.
Add filters, envelopes and modern tricks such as warping and drawing your own tables, and you get the defining sound of modern bass music: evolving pads, growling basses and leads that move as you hold the note.
In software: Xfer Serum, Vital (free) and u-he Hive 2.
Granular synthesis
Granular synthesis chops a sound into tiny grains, usually a few milliseconds to a few hundred milliseconds long, and plays them back in clouds. Change the grain size, density, pitch, position and randomness, and a short recording can become a shimmering pad, a frozen drone or a stuttering glitch.
The ideas go back to composer Iannis Xenakis and were developed in computer music by Curtis Roads, but it only became practical in real time with modern computers. It's at its best for ambient textures, cinematic soundscapes and sound design.
In software: Baby Audio Grainferno and, as an effect, Sugar Bytes Graindad.
Physical modelling synthesis
Physical modelling uses maths to simulate how a real object vibrates: a string being plucked, a tube being blown, a membrane being struck. Rather than playing back a recording, it calculates the sound, so it responds naturally to how hard and how long you play.
Yamaha's VL1 brought it to a commercial keyboard in the mid-1990s. Today it's used for realistic pianos and instruments, and increasingly for sounds that couldn't exist physically at all. It's expressive and organic, and great for plucks, mallets and tones that feel alive.
In software: Baby Audio Atoms, built on a mass-spring simulation, plus Pianoteq and AAS Chromaphone.
Sample-based synthesis
Sampling records real sounds and plays them back across the keyboard, with filters, envelopes and effects on top. The Fairlight CMI and E-mu Emulator brought it to studios around the turn of the 1980s, and samplers like the Akai MPC went on to shape hip hop. Romplers took the idea further, with large libraries of pre-recorded instruments ready to play.
It's the most realistic way to reproduce acoustic instruments, and in genres built on flips and chops, sampling is the whole point.
In software: TAL-Sampler, which also emulates the converters of classic hardware samplers, and Serato Sample.
Hybrid synths
Most modern synths don't stick to one method. u-he Zebra 3 combines several oscillator types in a modular layout, and the big all-rounders mix wavetables, samples, FM and granular engines in one instrument. Once you understand the building blocks, those synths become much easier to get your head around.
Which one should you learn first?
Subtractive. It's the foundation for almost every other type, the controls make immediate sense, and most presets on every synth still use a filter and an envelope somewhere. Once you're comfortable there, try wavetable, then FM. Have we missed anything?
If you'd like something to practise on for free, our guide to the best free synth plugins has a synth for almost every type of synthesis here.