It is based on the same principle as sampling. However, the samples are not played back conventionally, but are instead split into small pieces of around 1 to 50ms. These small pieces are called grains. Multiple grains may be layered on top of each other, and may play at different speeds, phases, volume, and frequency.

At low speeds of playback, the result is a soundscape, often a cloud, that is manipulable in a manner unlike that for natural sound sampling or other synthesis techniques. At high speeds, the result is heard as a note or notes of a novel timbre. By varying the waveform, envelope, duration, spatial position, and density of the grains, many different sounds can be produced.