Research Cell: An International Journal of Engineering Sciences

Akshat Arora ¹, Nikhil Panwar ¹, Pankaj Sikka ¹

Affiliations
1 Jaypee Institute of Information Technology, Noida, IN

Abstract

Increase in the amount of digital audio media in recent years, demands for content based retrieval of multimedia. However India classical Music is untouched from this revolution. Hence, this paper presents a comparative study on content based identification and retrieval for Indian Classical Music. The paper focuses on comparative implementation of various features like mfcc, bark scale based features etc for reliable, fast and robust identification of Indian classical music which can be run on the resources provided by today's standard computing platforms. Besides, a new modified feature has also been proposed to minimize the drawbacks (i.e. space and time complexity) of already existing features. The algorithms are tested on Indian Classical Music, which exhibits certain characteristics that make the identification task harder when compared to Western music.

Keywords

Content Based Audio Retrieval, Indian Classical Music, Audio Fingerprinting, MFCC, FFT.

Full Text