OSCAR uses Markov chains to generate musical material in accompaniment with the music being played by a human musician. The basics of Markov chains are discussed in detail in the next section; OSCAR is an exercise in their application. The following rough sketch of OSCAR is intended only as an introduction, which will be filled in with great detail in future sections.

OSCAR is comprised of three major components, called Hands, Ears, and Brains. The Ears analyzes music before a performance, and, from that analysis, constructs transition probability arrays that describe the music that has been analyzed. These arrays are stored in the Brains. During a performance, the Hands accesses the information in the Brains and uses it to construct musical material to accompamy a human musician.

A great number of files make up each of these three sections. A listing of OSCAR's files, along with a brief description of the function of each file, is included here.

A general understanding of Markov chains is needed before one can understand OSCAR. An introduction to Markov chains, therefore, begins the next section.