Poetic meter classification using TMS320C6713 DSK

Abstract

Poems are the vital part of literature which communicates through rhythm and its apparent meaning. Meter is a set of well defined rules that gives rhythm to the poetry. In this paper, we present a poetic meter classification algorithm implemented on TMS320C6713 floating point digital signal processor. Meter classification is done based on the similarity measure obtained from the matching of melodic pitch of poems. At first, melodic pitch is computed from the audio files using a Sonic Visualizer plug-in called MELODIA and then classification is performed using dynamic time warping (DTW) framework. The systematic evaluation is done on subset of a new database, created on studio environment, comprises of 10 meters of Malayalam language. The algorithm is successfully implemented on C6713 DSK. The results demonstrate the potential of the algorithm in music information retrieval applications.