Can a computer be used to analyse the mood and genre of different pieces of music and so offer an insight into how influential a piece might be, might it even be used to write the next mournful classical movement, angst-ridden emo rock song, or bouncy, hooky, catchy one-hit wonder?
Research International Journal of Arts and Technology from a team in China demonstrates how a directed weighted complex network and statistical methods can be used to carry out an evaluational analysis. Xinyan Ma, Xinyu Zhou, and Tingting Mo of the School of Mathematics and Information Science at Guangxi University in Nanning have used their knowledge of graph theory and cluster analysis in their work. Through this, they can observe trends of musical development among artists and genres.
One might suggest that given the perhaps entirely subjective and emotional responses we have to music that a mathematical analysis would not be possible. But, the underlying forms and formalities of music, its structures and styles, are in many ways mathematical and can be teased apart to reveal insights that perhaps go beyond our emotional response. Moreover, the tools of machine learning and artificial intelligence might even be taught to extend this kind of analysis to allow us to extract or even recreate particular musical notions from music. An objective means of music analysis could also be useful in studying how specific pieces of music and genres affect culture over time.
The network models generated by the team offer a visual way to look at music in terms of its various characteristics and how these correlated historically with popularity and the place of that music within the cultural environment. Ultimately, the team writes, “this project can help music lovers to further understand different types of music.” But, the findings offer much more than that, pointing to a deeper appreciation of different musical genres. The work might also be useful in priming music recommendation tools and so open up listeners to the experience of artists and genres they may not have heard before but find that they enjoy listening to.
Ma, X., Zhou, X. and Mo, T. (2021) ‘Evaluation analysis of music based on directed weighted complex network and statistics’, Int. J. Arts and Technology, Vol. 13, No. 4, pp.315–335.