Editorial: Music and AI

Abstract

Computational techniques have been used in a variety of ways for the creation and production of musical compositions with this field predating even digital music synthesis. Algorithmic music composition techniques range from the use of stochastic processes to create music based on random events to learning-based approaches. The paper “Computational Creativity and Music Generation Systems: an Introduction to the State of the Art” (Carnovalini and Roda) reviews work extending over six decades, organizing their presentation by methods, ranging from Markov chains to deep networks, and offering a set of open challenges. An extensive bibliography is included. Just as the techniques used to generate the music are varied, so too is the style of music generated, from the creation of written scores to musical accompaniment. “Evolving Musical Sight Reading Exercises Using Expert Models” (Pierce et al.) presents a novel evolutionary algorithm for generating monophonic sight-reading exercises in the Western art music tradition. Drawing on expert models of published sight-reading exercises, the evolutionary process draws on six fitness measures to create new exercises designed for specific grade levels of musical instruction. These include target note lengths, target rest lengths, allowable lengths, target intervals, allowable intervals, and melody shape. “On the Adaptability of Recurrent Neural Networks for Real-Time Jazz Improvisation Accompaniment” (Kritsis et al.) describes the basic implementation of an artificial jazz accompanist system that provides real-time accompaniment to a human musician soloist, based on a given harmonic description of lead sheet chord symbols. Recurrent Neural Networks are employed both for modeling the predictions of the artificial agent and for modeling the expectations of human intention. Fuzzy logic is a branch of AI that is often overlooked in this age of big data and neural networks. Nevertheless, fuzzy logic can be a powerful tool for modeling and learning music information expressed as signals, parameters or symbols. Fuzzy logic offers a useful framework for expressing models that can assist learning from less data. “Creating Music with Fuzzy logic” (Cadiz) offers an introduction to this field and describes a software toolkit created for composition and real-time music applications. Edited and reviewed by: Sriraam Natarajan, The University of Texas at Dallas, United States