Granular synthesis involves the use of short, segmented events of sound of around 5 to 50 milliseconds each (known as "grains" or “clouds”) that are manipulated and then subsequently pieced back together in order to form new complex sounds. These new sounds are typically pad-like but can be used for melodic parts as well. The "grains" themselves can be derived from a wavetable, FM synthesis, and sampled sounds. They are typically short to avoid the perception of pitch and can also be modified using parameters such as grain duration, grain density, and grain spacing. The sonic components can also be altered by their resulting in more complex sounds.