Our proposed approach detects music structures by looking at beat space segmentation, chords, singing-voice boundaries, and melody- and content-based similarity regions. Experiments illustrate that the proposed approach is capable of extracting useful information for music applications.
Index Terms:
segmentation, voice boundaries, content-based, octaves, subbands
Citation:
Namunu C. Maddage, "Automatic Structure Detection for Popular Music," IEEE MultiMedia, vol. 13, no. 1, pp. 65-77, Jan.-Mar. 2006, doi:10.1109/MMUL.2006.3