Tendencies, perspectives, and opportunities of musical audio-mining
Marc Leman, Lieven Clarisse, Bernard De Baets, Hans De Meyer, Micheline Lesaffre, Gaétan Martens, Jean‐Pierre Martens, Dirk Van Steelant · Ghent University Academic Bibliography (Ghent University) · 2002
Content-based music information retrieval and associated data-mining opens a number of perspectives for music industry and related multimedia commercial activities. Due to the great variability of musical audio, its non-verbal basis, and its interconnected levels of description, musical audio-mining is a very complex research domain that involves efforts from musicology, signal processing, and statistical modeling. This paper gives a general critical overview of the state-of-theart followed by a discussion of musical audio-mining issues which are related to bottom-up processing (feature extraction), top-down processing (taxonomies and knowledge-driven processing), similarity matching, and user analysis and profiling.