Types of synthesis: subtractive, wavetable, FM, additive, granular and modular.

Each synthesis family answers the same question in a different way: how to achieve the timbre you want. Some start with too much material and cut it down; others add from zero; others play or scatter previously recorded audio. That is why two presets with the same note and filter can sound worlds apart: it is not the interface, but the engine underneath.

This guide organises the families you will find in manuals and shops, explains the sound each produces and what it is used for in practice. If you are still unsure what each synthesiser block does, first revisit the fundamentals of synthesis; if you want to know which particular plugin offers each family, keep reading and then move to the synthesiser comparison.

Which family makes which sound

The following table is the starting map: family, principle and use. After you read it, this guide’s sections break each family down into its signal chain and its natural territory within a production.

Subtractive and Wavetable: what you hear in most tracks

Subtractive is the family inherited from analogue: choose a wave with many harmonics (saw, square or a mixed shape) and remove the excess with a low-pass filter. All the creative work is in the filter, resonance and envelopes, which is exactly what you saw in the fundamentals. Its limitation is that the starting material changes little: without an envelope or modulation, the sound remains flat.

Wavetable keeps the shaping philosophy but expands the source: instead of one wave, there is a table with dozens of shapes, which you can move through while the note plays. That is why timbres evolve without the filter needing to move so much, and why modern bass sounds feel “alive” during sustain. Serum’s manual describes the instrument as a combination of wavetable, subtractive, granular and other techniques in one workflow, which is common today: the family is determined by whichever engine you use at that moment.

In practice, subtractive handles the classic bassline, the familiar lead and filtered pads; wavetable handles sounds needing a change of texture within the note and lines with rhythmic movement. It is also useful to know when each wins: if a sound depends on a filter sweep and a short envelope, subtractive gets there sooner with fewer parameters; if it depends on timbre changing while a note sustains, wavetable avoids extreme resonant settings and unusual mismatches. In modern hybrids, the boundary is a box: choose the engine by slot, and the rest of the chain stays the same.

FM, Additive and Phase distortion: timbres a filter cannot reach

FM (frequency modulation) uses one wave to alter the frequency of another. With integer frequency ratios and little modulation, it produces rounded bell tones and very defined bass; increasing the index brings out metallic sounds, percussion and digital distortion. It is a family with few controls and results that are hard to predict, mastered through guided attempts rather than formulas.

Additive does the opposite of all the others: it adds partials, one by one, until the timbre is reconstructed. It is extremely clean and controllable but costly, because every partial needs its own generator. Organs and bells are its territory, and modern additive engines bring it closer to timbral evolutions impossible to achieve with a filter.

Phase distortion occupies the gap between the two. As Arturia documents in its Casio CZ emulation, while FM modulates one wave’s frequency with another, phase distortion alters the carrier’s time domain and achieves timbres from digital to analogue with very few controls. It sounds like artificial sawtooths from the eighties: bass, bright sounds and keyboard synthesisers that do not need an expensive filter to sound defined.

Granular, Physical modelling and Sampling: inside audio

Granular cuts material into millisecond micro-regions and scatters them in time, pitch and space. You can stretch a voice for a minute without changing its pitch or hearing where it came from, or turn a hit into a cloud. It is the family for evolving pads, backgrounds and transition FX; its risk is that, if you use grains everywhere, everything ends up sounding like the same texture playground.

Physical modelling does not simulate waves: it models the behaviour of an instrument, a struck string, a blown tube or a membrane. Its advantage is expressiveness: dynamics, strike position and noise respond as they do in the real instrument. Arturia calls it the Modal engine in Pigments and uses it for precisely this: organic sounds and modelled percussion that neither a filtered wave nor a sample can provide.

Sampling plays previously recorded material within the synthesis engine. It remains the fastest route to real sounds and percussion layers, and it is worth distinguishing it from synthesis: here you do not generate the timbre, but repeat it, trim it and process it.

Modular: wiring the chain yourself

In a fixed synthesiser, the route is already decided: oscillator, filter, amplifier and effects. In a Modular synthesiser, you choose the route with cables or software connections: the envelope can go to pitch, the LFO to resonance, the filter to volume and the result to another filter. The VCV Rack manual describes exactly this model: a virtual Eurorack studio in which modules are selected and connected one by one.

In exchange for that freedom, you pay two things: time and chaos. Most first patches end in silence or a loop you do not know how to undo. The classic recommendation is to reconstruct the standard route first (oscillators to mixer, mixer to filter, filter to amplifier), then move one connection at a time. Once you master that basis, modular stops being a maze and becomes the only way to make patches no library covers.

Family according to the instrument you need

With the complete map, the decision becomes mechanical:

• Focused bass: subtractive with one wave and a filter envelope; if you want evolution, wavetable. Remember that the sub remains an almost intact sine.
• A lead that slots straight in: subtractive with unison, or FM if you want a metallic edge.
• A substantial pad: wavetable or additive for the body, granular for the tail and some phase distortion if you want an eighties character.
• Percussion and hits: FM and physical modelling provide the transient; sampling handles what already exists.
• FX and transitions: granular and filtered noise; time matters more than tuning here.
• A patch nobody has made: modular, and patience.

Choosing without installing ten synthesisers

Three filters, in this order:

1. Which engine you need for this sound. If the answer is “a filtered wave”, any subtractive engine works; if it is “evolution within the note”, you want wavetable; if it is “texture from a sample”, granular.
2. How many families you want in one window. Modern hybrids stack engines and avoid jumping between plugins, but also multiply parameters: more power, more friction.
3. How much CPU you can spend. Granular, physical modelling and unison with many voices cost more than one wave with a filter. In projects with many tracks, the chosen engine is also a mixing decision.

The trap is choosing by the number of families in the brochure. The right family is the one that solves the sound in front of you in the fewest steps; the rest is legitimate curiosity, but for another day.

A short method to avoid losing the afternoon: describe the sound in a sentence (“a bass that opens on every hit and closes at the end”), translate that sentence into blocks (rich wave, filter with a zero-attack envelope), and only then open the synthesiser. If you cannot form the sentence, no family will decide for you, and trying five engines with the same vague idea only multiplies the doubts.

Next step

You now know what each family does. The next step is to place the instruments that implement them on the map: Serum, Vital, Pigments, Sylenth1, Diva and others, with their strengths and actual costs in use. That comparison is in Synthesisers: Serum, Vital, Pigments, Sylenth1, Diva.