What Granular Synthesis Actually Is
Most explanations of granular synthesis start with the metaphor and skip the mechanism. The metaphor is: you take a sound and shred it into tiny pieces, then reassemble those pieces in new ways. That's accurate but it doesn't tell you why it sounds the way it does, or why the destruction of the original is the point rather than a side effect.
Here's the mechanism: granular synthesis works by slicing an audio sample into very short segments — grains — typically between 1 and 100 milliseconds long. Each grain is then played back with its own envelope (attack, sustain, release), pitch, and position in the stereo field. The synthesizer plays many of these grains simultaneously, overlapping them, and the result is a texture rather than a sound.
The original sample is still in there, technically. But it's been distributed across time and space in a way that makes it unrecognizable. That's not a bug. That's the instrument.
Why the Destruction Is the Point
When you record a piano note, you capture a specific event: a hammer hitting a string at a specific moment, in a specific room, with a specific decay. That event has a beginning, a middle, and an end. It exists in time.
Granular synthesis removes the event from time. It takes the sonic material of the piano note and spreads it across a texture that can sustain indefinitely, that can be pitched without changing its duration, that can be reversed without sounding reversed. The piano note becomes a cloud of piano-ness.
This is why granular synthesis sounds the way it does — that characteristic shimmer, that sense of something familiar made strange. You're hearing the material of a sound without its temporal structure. The structure was what made it recognizable. Without it, you're left with something that feels like a memory of the original rather than the original itself.
There's a connection here to fringe signal theory — the idea that information and its carrier are separable, and that separating them reveals something about both. Granular synthesis is a demonstration of that principle in audio.
Practical Parameters
The main controls in a granular synthesizer are grain size, grain density, position, and randomization. Grain size determines how much of the original sample each grain contains — smaller grains produce more abstract textures, larger grains retain more of the original's character. Density controls how many grains are playing simultaneously. Position determines where in the sample the grains are drawn from. Randomization adds variation to all of these parameters.
The interesting territory is usually in the extremes. Very small grains (under 10ms) produce pitched noise — the sample's pitch information is preserved but its timbre becomes almost unrecognizable. Very large grains (over 100ms) start to sound like a stuttering version of the original. The sweet spot for most musical applications is somewhere in the 20-50ms range, where you get texture without losing all connection to the source material.
What It's Good For
Granular synthesis is particularly useful for creating sustained textures from percussive sounds, for pitch-shifting without time-stretching artifacts, and for creating that specific quality of sound that feels like it's dissolving at the edges. It's also useful for creating transitions — granularizing a sound at the end of a section and letting it dissolve into the next section is a technique that works in almost any genre.
The music and culture around experimental production has been using granular synthesis for decades, but it's become more accessible as the tools have improved. Most modern DAWs have at least one granular plugin, and several hardware synthesizers have granular engines built in.
The thing to remember is that granular synthesis is not a way to make sounds better. It's a way to make sounds different — specifically, to make them exist in a different relationship to time. Whether that's useful depends entirely on what you're trying to do.
More on the maker's process and at endof8.