Applying text analytics on music emotion recognition

Abstract

Music is most important part in human life. When different kind of words are prepared a new sound which is enjoyable to the human beings, it is called music. Music is not just a source of entertainment. It is something more than entertainment. Title should come after the song is finished and should reproduce a termination of the lyrical content. In this paper our proposed system have recommend a proper song title based on song lyrics. We have applied topic model algorithm Latent Dirichlet allocation (LDA) for song title recommendation. In this paper another experiment is music emotion recognition. Song feel emotionally different to listeners depending on their lyrical contents. Emotions classify is so difficult through the existing music emotion classification method .We have extracted eight features from song lyrics. We propose a method for lyrics based emotion classification using feature selection. We also proposed another experiment music personality trait. We have to generate a customize dataset based on music interest and 20 questions of big five personality model. Our proposed module would be helpful for user. Song title recommendation system produces satisfactory result. We may use this module to recommend song title from lyrics. Music emotion recognition system will help to predict the overall emotional state of a user. Music personality traits could be useful to find out the personality measurements of any user.