Visualization of Similarity Distribution of Lyric Collections

Lyric is an important factor in music. Many pop music listeners have their favorite lyrics. It is an interesting task to find similar lyrics in pop song collections. Visualization is an effective tool for discovering interesting trends in lyrics from such song collections. This chapter presents studies of visualizations of lyrics collections applying LDA (Latent Dirichlet Allocation) to calculate features of lyrics and dimensionality reduction to represent the distribution of the features as scatterplots. The first study applies guided LDA to interactively consume guided words and adjust the visualization results from users’ viewpoints. The second study analyzes the differences in lyrics based on the gender of singers.