hide
Free keywords:
music information retrieval; open dataset; Indian music; Carnatic; Hindustani
Abstract:
This commentary explores features of the "Saraga" article and open dataset, discussing some of the issues arising. I argue that the CompMusic project and this resulting dataset are impressive for their sensitivity to cultural specificities of the Hindustani and Carnatic musical styles; for example, the dataset includes manual annotations based on music theoretical concepts from within the styles, rather than imposing conceptual categories from outside. However, I propose there are aspects of the dataset's manual annotations that require clarification in order for them to be used as ground truths by other researchers. In addition, I raise questions regarding the representativeness of the dataset – an issue that has ethical implications.
