Music genre classification using spatan 6 FPGA and TMS320C6713 DSK

Abstract

Music genre classification is one of the most important element of the music information retrieval (MIR) community. In this paper, we present a music genre classification system using field programmable gate arrays (FPGA) and dedicated DSP processors. The proposed system uses FPGA based acoustic feature extraction of mel frequency cepstral coefficients (MFCC) and dynamic time warping (DTW) based classifier using TMS320C6713 floating point processor. We successfully implemented MFCC extraction algorithm on Spartan 6 FPGA clocked at 150 MHz with support from TMS320C6713 floating point processor followed by DTW based matching engine. The paper attempts to implement music genre classification algorithm in hardware, yielding competitive performance in music information retrieval applications.