A grain is a short fragment read from a buffer, and it is the basic unit of granular synthesis

Granular synthesis builds textures by generating and layering many very short audio fragments called grains, each typically a few milliseconds to a few hundred milliseconds long. The starting point is a sound file loaded into a buffer; a grain is produced by reading an arbitrary sub-region of that buffer rather than playing the file end to end. In Max/MSP you load a file into a buffer~ (e.g. via a drop-file object) and read a grain with a playback object by specifying a start position and a duration — for example reading 200 milliseconds beginning at one point, then reading another 200 ms from a different offset. The key insight distinguishing granular from ordinary sample playback is that you treat the buffer as a store to select tiny windows from anywhere within it, and repeat those reads, instead of playing it linearly.

Examples

In Max/MSP load a recording into a buffer~, then read a 200 ms grain from the start; change the start offset (e.g. to 400 ms in) to take the next grain from a different point in the file.

Assessment

Explain what distinguishes a grain from ordinary sample playback, and describe how you would read one grain from a specific position in a buffer.