Synthesis of sound textures, such as rain, wind or crowd applause, is an important and useful application in many domains such as the movies industry, multimedia, video games, music and more.
The necessity for a flexible and efficient synthesis method which can produce a natural sound is obvious, yet, up to this day, no automatic synthesis method is found in common use.
There are few different approaches towards sound texture synthesis. In this project we will discuss a method established on a data base. The method we have implemented is based on the fact that sound textures are composed of micro-events which have time-constant or cyclical statistical features that form atoms – bundles of micro events. The atoms enable us to extract all of the texture's data in finite and mostly short time.
The project shows a product development process – from an idea to an algorithm and a user interface.