音楽ストリーミングサービスで配信された楽曲の特徴量を用いた音楽嗜好性の国際比較

抄録

In recent years, the environment surrounding the international music business has been changing rapidly. In order to succeed in the international music distribution business, it is necessary to understand the heterogeneity of music preferences among nations. In this paper, we investigate music preferences in five countries: the United States, Germany, the United Kingdom, France, and Japan by using big data from music streaming services. We employ quantitative music features, which are different from qualitative approaches, to find differences in music preferences of the top 200 best-selling songs for the five countries from 2017 to 2020. The differences among the five countries are visualized using the INDSCAL method, which is a kind of multidimensional scaling method. The empirical relative frequency distributions of the features of music consumed in each country were constructed as feature vectors and applied to the INDSCAL method. As a result, it was found that (i) the music preferences of the five countries are heterogeneous in terms of music features, (ii) the UK has average music preferences, (iii) Japan and the other four countries have different music preferences, and (iv) France and the other four countries also have different music preferences. The results can be applied to the realization of better music recommendation systems and marketing in the music business.