Music genres are essential for organizing and categorizing music, making it easier for listeners to discover, enjoy, and connect with styles that resonate with them. Genres also carry historical, cultural, and sonic significance. Playlists, which often focus on a single subgenre, have become an increasingly popular way to discover new music.

We address the multi-label classification problem to identify a song's genre(s) using acoustic features extracted from audio files. We train a variety of supervised learning models to determine genre. Rather than focusing on broad genres (e.g., jazz, hip hop, electronic), we concentrate on four subgenres of electronic music: techno, house, trance, and drum and bass. While these subgenres are distinct and well-defined, they can be challenging to differentiate. We train various models, including XGBoost, and neural networks on data obtained from AcousticBrainz.