Music is and has been an integral part of our society since time immemorial. It is a subtle display of a person’s emotions. Over the decades even though the way music is composed or heard has greatly evolved but what has remained constant is the entwined relationship it shares with mood. The kind of music one listens is to be governed solely by their mood at that instant. This paper proposes an automated and efficient method of classifying music on the basis of the mood it depicts, by extracting suitable features that show significant variation across songs. A database of 300 popular Bollywood songs was taken into consideration in which timbral and temporal features were extracted to classify songs into four moods: Happy, sad, relaxed and romantic. 200 songs were used to train the model by using Multilayer perceptron with backpropagation algorithm. The model exhibited an accuracy of 75% when tested over a set of 100 songs.
CITATION STYLE
Tyagi, P., Mehrotra, A., Sharma, S., & Kumar, S. (2016). Audio pattern recognition and mood detection system. In Advances in Intelligent Systems and Computing (Vol. 436, pp. 321–332). Springer Verlag. https://doi.org/10.1007/978-981-10-0448-3_26
Mendeley helps you to discover research relevant for your work.