Composing Music with Neural Networks and Probabilistic Finite-State Machines

摘要

In this paper, biological (human) music composition systems based on Time Delay Neural Networks and Ward Nets and on a probabilistic Finite-State Machine will be presented. The systems acquire musical knowledge by inductive learning and are able to produce complete musical scores for multiple instruments and actual music in the MIDI format. The quality of our approaches is analyzed in objective and subjective manner with existing techniques.