Staring at a sample and wanting something stranger than a loop? Granular synthesis slices a sound into tiny grains and replays them in huge numbers, turning any recording into a pad, a cloud or a shimmer.
| Item | Value |
|---|---|
| Grain duration in Truax’s description | Less than 50 ms |
| Typical grain duration | 10-30 ms |
| Gabor’s “quantum of sound” idea | Proposed in 1947 |
| Granular synthesis first suggested for computers | Xenakis (1971), Roads (1978) |
| Real-time implementation by Barry Truax | 1986 |
Figures come from Barry Truax’s granular synthesis page at Simon Fraser University. Historical dates and typical ranges can change with new scholarship, so check the source before you rely on them.
What granular synthesis actually does
A grain is a very short slice of audio, usually cut from a sample. Heard alone, it is the merest click or point of sound, as Truax puts it.
The magic appears when you play many grains per second and let them overlap. Their sum becomes a continuous texture whose character depends on where the grains come from and how they are shaped.
Think of it like a mosaic. Each tile is almost meaningless on its own, yet the picture changes completely depending on which tiles you pick, how closely you pack them and how much you shuffle them.
That is why a single vocal phrase can become a choir-like pad, a rattling swarm or a frozen chord. The source supplies the colour; the grain settings supply the shape.
This is a different idea from looping a sample or playing back a cycle of a waveform. If you want the contrast, wavetable synthesis scans through stored single-cycle shapes, while granular engines scatter and stack fragments of real recordings. Both belong to the wider craft of sound design.
A short history of the grain
The physicist Dennis Gabor proposed the grain as a quantum of sound in 1947, an indivisible unit of information from the psychoacoustic point of view. Everything larger, in his thinking, is built from such units.
Granular synthesis was first suggested as a computer music technique by Iannis Xenakis in 1971 and Curtis Roads in 1978. Barry Truax then implemented it in real time in 1986, using a digital signal processor controlled by a microcomputer.
Today the same idea sits inside software samplers, effects and standalone instruments, so you no longer need a research lab to try it.
The four controls that shape every grain cloud
Most granular instruments expose the same handful of knobs. Learn these and you can move between plugins easily.
- Grain size: how long each grain lasts. Changing grain length can make the original recording easier or harder to recognise.
- Density: how many grains play per second. Low density gives isolated clicks and stutters; high density fills the space with a continuous texture.
- Position: where in the source audio grains are pulled from. Freezing it loops one tiny fragment; automating it slowly scrubs through the material.
- Spray: a random offset applied to the read position, so grains come from a range around the chosen point rather than one exact spot.
Many instruments let you go longer than the classic grain lengths in the table above, which is where the source starts to become recognisable again.
Source for the control definitions: Beat Kitchen’s granular synthesis chapter.
Why the grain envelope matters
Every grain needs a fade in and a fade out. Truax calls the grain envelope the key to all granular techniques: brief linear fades at the edges of sampled grains help avoid introducing clicks.
The slope also colours the result. Sharper attacks produce broader bandwidths, much as very short grain durations do.
In practice this gives you a simple rule of thumb:
- Soft, rounded windows blend into smooth pads.
- Steeper windows keep more bite and suit rhythmic or glitchy material.
The same thinking applies to any shaped sound. If you have used an adsr envelope on a synth voice, a grain window is a miniature version of that attack and decay, repeated over and over.
Because the window is so short, you rarely hear it as a shape. You hear it as tone colour: whether the cloud feels smooth and breathy or sharp and clicky.
When a texture sounds harsh, check the window before anything else. A gentler fade often fixes the problem without touching size or density.
Synchronous and asynchronous clouds
How grains are spaced in time changes the sound as much as their size. Truax lists several common types.
- Quasi-synchronous: a stream of equal-duration grains at regular intervals. With grain durations less than 50 ms, this produces amplitude modulation, so you hear a buzzy or pitched quality.
- Asynchronous: grains are scattered stochastically, with timing set by density. The result is a diffuse cloud with no obvious pulse.
If a patch sounds too mechanical, loosen the timing or raise spray. If it sounds too messy, tighten both and shorten the spread.
Practical ways to start
Choose a source with movement: a spoken phrase, a bowed string, a field recording. Even static sources can produce evolving clouds when grain parameters change over time.
- Set a short grain size and moderate density, then listen to the raw cloud.
- Freeze position on an interesting moment to get a sustained drone.
- Add a little spray until the repetition softens.
- Automate position slowly so the texture evolves over time.
- Lengthen the grains if you want the source to creep back into focus.
Make one change at a time. Size and density interact, so moving both together makes it hard to learn what each is doing.
Sound design ideas to try
Once the controls feel familiar, granular tools become a source of ideas rather than a single effect. These are starting points, not rules.
- Frozen pads: hold position on a vowel or a plucked note and let spray add slow movement. This can create a sustained texture, with spray helping to reduce obvious repetition.
- Textural beds: use a field recording such as rain or traffic, with long overlapping grains, to build an ambient layer under a track.
- Stutters and glitches: lower density so individual grains poke through silence, then use a sharper window for rhythmic bite.
- Risers and transitions: automate position and size together so a sound smears and dissolves into a wash.
- Vocal smears: scrub position slowly through a spoken line to stretch it without a simple repeat.
Layering helps too. A granular cloud under a clean, unprocessed version of the same sample often sounds richer than either alone.
Finally, record your settings as you go. Small moves in size, density and spray can change a patch dramatically, and a note about what worked saves a lot of guessing later.
Frequently asked questions
How long is a grain?
Show answer
Truax gives less than 50 ms, typically 10-30 ms. Plugins often allow longer settings for more recognisable results.
Do I need a special synth to use granular synthesis?
Show answer
No. Many samplers and effects include a granular mode, and any tool exposing size, density, position and spray will do the job.
Why does my granular patch click?
Show answer
Grains without a fade at their edges add clicks. A short attack and decay on each grain’s envelope prevents this.
What does spray do?
Show answer
It randomises the read position within a range, so grains come from around a point rather than one exact spot.
