International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395 -0056 Volume: 04 Issue: 04 | Apr -2017 www.irjet.net p-ISSN: 2395-0072 © 2017, IRJET | Impact Factor value: 5.181 | ISO 9001:2008 Certified Journal | Page 3412 MUSIC GENRE CLASSIFICATION Ayush Kumar 1 , Bojja Krishna Sai Siva 2 , G Sumin Reddy 3 , M.R.Rashmi 4 1 Ayush Kumar, Department of Computer Science, The National Institute of Engineering, Mysore, Karnataka , India 2 Bojja Krishna Sai Siva, Department of Computer Science, The National Institute of Engineering, Mysore, Karnataka , India 3 G Sumin Reddy, Department of Computer Science, The National Institute of Engineering, Mysore, Karnataka , India 4 M.R.Rashmi, Assistant Professor, Department of Computer Science, The National Institute of Engineering, Mysore, Karnataka , India ---------------------------------------------------------------------***--------------------------------------------------------------------- Abstract – In this paper , we have put forth a music genre classification approach based on Mel Frequency Cepstral Coefficients (MFCC) . The Mel Frequency Cepstrum (MFC) encodes the power spectrum of a sound. It is calculated as the Fourier transform of the logarithm of the signal's spectrum. A modulation spectrogram is used corresponding to the collection of modulation spectra of Mel Frequency Cepstral Coefficients (MFCC) will be constructed. A spectrogram is a visual representation of the frequency content in a song. It shows the intensity of the frequencies on the y axis in the specified time intervals on the x axis; that is, the darker the color, the stronger the frequency is in the particular time window of the song. Each genre existing as of now is carefully designed with a particular number of prototype vectors which were designed with appropriate algorithms. An information fusion approach comprising of both feature and decision level fusion is involved for the appropriate outcome. Recently there was a development which saw a million song dataset being released which poised all the song features and its metadata. But , we find that our method comprising of the salient features listed above improves the accuracy for detecting genre of a music file . Key Words: Genre Classification, Mel Frequency Cepstral Coefficient, Mel Frequency Cepstrum, Modulation Spectrogram . 1.INTRODUCTION The first thing that strikes one’s mind with the heading is that what is a music genre? Music genre can be defined as a category or rather conventional category that recognises the characteristics or traits of sub-division of the music file belonging to a traditional or any conventional established music form The term Music Genre Classification can be explained as categorising of music samples. A music genre classifier plays a vital role in adjudging song samples in a preliminary stage, for instance if a fresh song has been recorded it will help in categorising the song into its conventional category. To determine the genre of a song it has to be distinguished by its unique audio features so that its contents can be analysed with respect to the produced wave signals. Another important aspect is the recognition of the instruments used in the song which is also known as the timbral characteristics. This plays a very vital role in specifying the music type based on the type of instrument helps us denote the traditional connect to the music. In this paper, we have used the Mel Frequency Cepstrum Coefficient (MFCC) for encoding the power spectrum of the sound with the calculation of the Fourier transform of the logarithm of the signal’s spectrum. Another important role player here is the spectrogram which helps us in the visual representation frequency content in a song. It shows the intensity of the frequencies on the y axis compared to the time interval on the x axis. This makes the job easier in predicting the potential genre of the music file thus fulfilling our search . 1.1 Proposed System The Proposed system makes use of Mel Frequency Cepstral Coefficients (MFCC) and spectrogram, thus to illustrate it further the system is divided into two steps as follows : i. Ceps construction phase ii. Genre classification phase i. Ceps construction phase- In this phase we use a python script which helps us to analyze and convert each file from the data-set in a representation that can be used by the classifier and be easily cached on to the disk. This little step prevents the classifier to convert the dataset each time the system is run.