A systematic review of artificial intelligence-based music generation: Scope, applications, and future trends
Miguel Civit, Javier Civit-Masot, Francisco Cuadrado, María José Escalona
Abstract
Currently available reviews in the area of artificial intelligence-based music generation do not provide a wide range of publications and are usually centered around comparing very specific topics between a very limited range of solutions. Best surveys available in the field are bibliography sections of some papers and books which lack a systematic approach and limit their scope to only handpicked examples In this work, we analyze the scope and trends of the research on artificial intelligence-based music generation by performing a systematic review of the available publications in the field using the Prisma methodology. Furthermore, we discuss the possible implementations and accessibility of a set of currently available AI solutions, as aids to musical composition. Our research shows how publications are being distributed globally according to many characteristics, which provides a clear picture of the situation of this technology. \n \nThrough our research it becomes clear that the interest of both musicians and computer scientists in AI-based automatic music generation has increased significantly in the last few years with an increasing participation of mayor companies in the field whose works we analyze. We discuss several generation architectures, both from a technical and a musical point of view and we highlight various areas were further research is needed.