This dataset contains multitrack audio recordings of rock and pop songs, along with structural segmentation ground truth annotations. In some cases the audio itself is provided, in others, we provide a hyperlink to commercial audio recordings. The format of the ground truth annotations conforms to the `Structural Analysis of Large Amounts of Music Information‘ (SALAMI) guidelines produced at McGill University.

For citations, please use this reference:

Hargreaves, S.; Klapuri, A.; Sandler, M., "Structural Segmentation of Multitrack Audio," IEEE Transactions on Audio, Speech, and Language Processing, vol.20, no.10, pp.2637-2647, Dec. 2012. http://dx.doi.org/10.1109/TASL.2012.2209419 URL: http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=6243190&isnumber=6313477