Markov chains for music generation

The thesis deals with the use of Markov chains in an algorithmic composition. It presents the theory of Markov chains including examples of common practical applications and the required music theory. The thesis also reviews methods used in algorithmic compositions, such as cellular automata, neural networks or evolutionary algorithms. The overview includes Markov chains too, which are described in a basic way that can be used to construct a model, whose parameters are determined by familiar musical compositions and thus can generate new works of music. The thesis further extends the way of the model construction by considering two parallel chains, one to do with tone lengths and the other with their pitch. The model demonstrates a different approach to creating states of these chains, along with an extended definition of the Markov chain. Such a definition differs from the basic one in what we consider a chain, in which the next state does not only depend on the current state but also on a certain number of previous ones. In the last part of the thesis, the model is implemented in the Python programming language.