Stochastic sound

Some ways of generating sound, often referred to as stochastic systems[18] use random or chaotic data. They may filter random data to obtain order through the mathematical rules of statistics and distribution. Alternatively they may use algorithmic[26] methods (see below) to generate quasi-random data with a high degree of complexity but well defined distribution. In most stochastic systems the generative part is purely random noise and the distribution is determined by how it is filtered. This follows the subtractive model from synthesis, where we start with a lot of data and selectively throw some of it away to get bands of activity where we want. In generate and test systems a filter can have a very complex bunch rules to see whether a chunk of randomly generated data should be passed or discarded. Distributions of data are often named according to recognised patterns from statistical theory such as uniform, linear, exponential and Gaussian. Each has particular applications in composition, such as determining note lengths and melodic density [26],[34], or in synthesis for approximating rainfall, city traffic noise and textures for footsteps[10]. Stochastic sound may be generative or interactive since user input can be applied to parameters of the generating equation or to subsequent filters that operate on generated data.