Proceedings of Computer Music Multidisciplinary Research
Abstract:
The top 40 chart is a popular resource used by listeners to select and
purchase music. Previous work on automatic hit song prediction focused on
Western pop music. However, pop songs from different parts of the world
exhibit significant differences. We performed experiments on hit song
prediction using 40 weeks of data from Chinese and UK pop music charts.
We used a set of ten common audio features with a time-weighted linear regression
model and a support vector machine model to predict whether a new song will
be a top hit or a non-hit.
Then we report on the features that performed best for predict
ing hit songs for both the Chinese and UK pop charts.
Our results indicate that Chinese hit song prediction is more accurate than the UK version of the experiment. We conclude that the audio f
eature characteri stics of Chinese hit songs are significantly different from those of
UK hit songs. The results of our work can be used to inform how music information retrieval systems are
designed for pop music from different musical cultures.