Detecting Melodic Motifs from Audio for Hindustani Classical Music

Melodic motifs form essential building blocks in Indian Classical music. The motifs, or key phrases, provide strong cues to the identity of the underlying raga in both Hindustani and Carnatic styles of Indian music. Thus the automatic detection of such recurring basic melodic shapes from audio is of relevance in music information retrieval. The extraction of melodic attributes from polyphonic audio and the variability inherent in the performance, which does not follow a predefined score, make the task particularly challenging. In this work, we consider the segmentation of selected melodic motifs from audio signals by computing similarity measures on time series of automatically detected pitch values. The methods are investigated in the context of detecting the signature phrase of Hindustani vocal music compositions (bandish) within and across performances.