[Review] An Introduction to Musical Metacreation


An Introduction to Musical Metacreation (Computers in Entertainment, Volume 16 Issue 2, Summer 2016)

Summary and Review

Artificial Intelligence (AI) has been successful in solving rational problems that have optimal or near-optimal solutions. Now then, can machines exceed their creator and become creative?

The difficulty innate in solving this problem is that creativity itself is difficult to define; there is no clearly accepted definition. A fairly moderate and uncontroversial definition by Boden is that creativity is the ability to come up with ideas or artifacts that are original and valuable. Also, the solution to problems that involve creativity usually lack optimality and are often vague. The end-point of the research is unclear since there is no best or clear solution for it. Nevertheless, they are clearly defined areas of activity that humans devote themselves to.

As a subfield of computational creativity, Musical Metacreation (MuMe) focuses on creating machines that are able to achieve creative musical tasks. Practitioners of the MuMe community build multiple MuMe systems for various purposes, but the main domain of the systems can be categorized as the following: Composition, Interpretation, Improvisation, and Accompaniment.

The author gives three additional points of discussion. First, when a machine generates some music, who is the author of the music? Is it the machine itself or the designer of the music? Also, if a corpus of musical works were given to the machine as a source of knowledge, should it be viewed as a collaborative authorship? Second, since MuMe is an inherently interdisciplinary field of study, the art-science dichotomy should be abandoned and the collaboration between scientists and musicians is important. Lastly, the MuMe community lacks standardization and iteroperability, since so many people participated without a structured guideline. The formation of an international organization providing structure and visibility to the field would be beneficial.

Appending my opinion, provided that the generated piece is different enough with the input musical works, the author should be the person who chose which work to input and how to set the parameters of the algorithm. Since most authors of a musical work did receive some kind of influence from previous musicians and their work, arguing that the authors of the input works have any right of the generated musical piece also implies that their own authorship of their work may not be exclusive.

Leave a comment