Song Similarity Analysis With Clustering Method On Korean Pop Song

Trend about the growth of music with the genre of pop, which is the most demanding genre by people, with the attraction that comes from the characteristics that make it unique, so that it becomes a factor in the popularity of the song. Since March 22, 2020, Korean songs or K-Pop have become a popular song genre, beating local pop music based on developments from Google Trends, many factors cause the popularity of the song genre, which comes from characteristics of the song. So these factors are sought through the song’s similarity. Due to the limitations of computers in predicting the song’s similarity-based on sound, so the model created to predict the song’s similarity using a grouping algorithm, those are KMeans Clustering and Self-Organized Map, began by learning the K-Pop songs characteristics obtained from Spotify until they were split into several groups. with each character that is suitable for each group, the factors of the characteristics of the K-pop song are obtained from the results of the analysis on each cluster of the K-pop songs. Thus, generally, the majority of K-pop songs are songs that are energetic, loud, and like to be danceable. Based on the results of the analysis by splitting them into 4 groups using K-Means because it is better than SOM, there are 2 pairs of clusters, each of which has similarities to each other in terms of tempo and mood descriptions. Where the factors of k-pop songs are represented by an explanation of the characteristics both in general and based on each cluster.