MUSIC ANALYZER AND PLAGIARISM

Milind Bhattacharya, Sweekar Bandkar, Amit Badala

Abstract: Music artists put a lot of effort in creating a perfect musical composition. However, some other artists directly lift the original pieces or change them marginally and use it under their own title to gain recognition and success. To curb these mounting violations in copyrights, we created an algorithm to compare pieces of music with various songs in a database to recognize plagiarism. The algorithm is based on the technique used by Shazam for music identification. The songs in the database are stored in the image form called acoustic fingerprints, which are derived from the spectrograms. A similar fingerprint is created for the song to be matched for plagiarism. Further, scatterplots are generated by comparing the fingerprint of the music under scrutiny to the fingerprints of all the songs in the database. Finally, all the scatterplots are subjected to a diagonal detection algorithm which returns the name of the song with highest percentage of match. This paper initially presents few theories on sound like amplitude and frequency, which forms the base to the concepts of spectrogram and acoustic fingerprinting. It also explains various techniques of image processing that are used in the creation of scatterplots and detection of diagonals. The MATLAB software is used for the explanation and implementation of the entire algorithm

Keywords: Music plagiarism, Music recognition, Acoustic fingerprinting, Spectrogram, Scatterplot, Diagonal detection

DOI: https://doi.org/10.15623/ijret.2014.0317002