Granular Synthesis is a method of synthesis that uses small pieces of audio in the range of 10 milliseconds to 1 second or something like that to generate sound "textures", implement effects such as pitch-preserving speeding or slowing of audio, and other effects. It is similar to [concatenative-synthesis], different in that [concatenative-synthesis] focuses on carefully connecting deliberately chosen samples in order to form very specific outputs such as [speech-synthesis], whereas granular synthesis often plays many overlapping "grains" which are sometimes chosen at random from different parts of a source sample.
See [granular-synthesizer]