A Study on Automatic Music Genre Classification Based on the Summarization of Music Data

Music summarization is often employed in music genre classification to reduce the computational cost. To examine how the accuracy of the music genre classification tasks is controlled with music summarization, this study compares our proposed technique that extracts summaries from multiple sections of the music data with a conventional technique that extracts a summary from a single section. The experimental result indicates that appropriate summarization of music data may potentially give better performances.