Sergio Ribes Miró

Degree in Sound and Image in Telecommunication

Polytechnic School. University of Alicante.

The algorithmic composition uses very diverse techniques to obtain musical works generated by formal processes. One of them is to compose sequences of notes based on the probabilities that a given note is preceded by another, a technique known as Markov Chain. These probabilities can be estimated from examples of compositions. More recent versions of the Csound system include tools whose use for this purpose have not yet been properly explored. This project focuses on studying these tools, by designing and implementing a system capable of composing a simple musical sequence using Markov chains, after having been trained using previous compositions. The designed system consists of two parts, the first one receives the MIDI files that the user wants to use as examples and 'trains' the system, generating the matrices that are used in the Markov process. The second part retrieves t


technical report (in Spanish):