Music Similarity and Retrieval

More about the book

This book summarizes diverse audio- and web-based approaches to music information retrieval (MIR) research, offering a holistic view by incorporating cultural and listener-centric aspects alongside traditional music signal processing. It covers methods that extract features from audio signals and contextual information, including cultural context from the web and user interactions with music. Following a document-centered paradigm, it discusses models of music similarity that derive computational features to represent music entities, such as songs, albums, or artists, and methods to calculate their similarities. Although not exhaustive in capturing all musical dimensions, these representations facilitate the discovery of music with similar qualities by providing abstract summaries of musical artifacts across various modalities. The text serves as a comprehensive and accessible introduction to music search, retrieval, and recommendation from an academic standpoint, making it suitable for newcomers to the field while highlighting the music domain's evolution within the broader information retrieval community. It is structured into four parts: Part I focuses on content-based MIR and feature extraction; Part II examines methods utilizing the cultural context of music; Part III discusses collaborative filtering and user-aware retrieval; and Part IV looks at current and future applications in music retrieval and recommendation.