Multimodal Mood Classification - A Case Study of Differences in Hindi and Western Songs

Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers, pages 1980–1989, Osaka, Japan, December 11-17 2016. Multimodal Mood Classification - A Case Study of Differences in Hindi and Western Songs Braja Gopal Patra, Dipankar Das, and Sivaji Bandyopadhyay Department of Computer Science and Engineering, Jadavpur University, Kolkata, India {brajagopal.cse,dipankar.dipnil2005,sivaji.cse.ju}@gmail.com Abstract Music information retrieval has emerged as a mainstream research area in the past two decades. Experiments on music mood classification have been performed mainly on Western music based on audio, lyrics and a combination of both. Unfortunately, due to the scarcity of digitalized resources, Indian music fares poorly in music mood retrieval research. In this paper, we identified the mood taxonomy and prepared multimodal mood annotated datasets for Hindi and Western songs. We identified important audio and lyric features using correlation based feature selection technique. Finally, we developed mood classification systems using Support Vector Machines and Feed Forward Neural Networks based on the features collected from audio, lyrics, and a combination of both. The best performing multimodal systems achieved F-measures of 75.1 and 83.5 for classifying the moods of the Hindi and Western songs respectively using Feed Forward Neural Networks. A comparative analysis indicates that the selected features work well for mood classification of the Western songs and produces better results as compared to the mood classification systems for Hindi songs. 1 Introduction Global digitization has led to music being available in the form of CDs, DVDs or other portable formats. With the rapid growth in Internet connectivity over the last decade, the audio or video music files are easily available and accessible over the World Wide Web. The number of music compositions created worldwide already exceeds a few millions and continues to grow. Similarly, the popularity of download- ing and purchasing of music from online music shops has also been increased at the same pace. Thus, the organization and management of the music files are the important issues to be tackled carefully. Re- cently, studies on music information retrieval (MIR) have shown that moods are desirable access keys to music repositories and collections (Hu and Downie, 2010a). In order to find out such access keys, most of the experiments on music mood classification of West- ern music have been performed based on the audio (Lu et al., 2006; Hu et al., 2008), lyrics (Zaanen and Kanters, 2010) and combination of both (Laurier et al., 2008; Hu and Downie, 2010b; Hu and Downie, 2010a). In case of the Indian music, few tasks have been performed on the Hindi music mood classi- fication based on the audio (Ujlambkar and Attar, 2012; Patra et al., 2013a; Patra et al., 2013b; Patra et al., 2016b), lyrics (Patra et al., 2015c) and combination of both (Patra et al., 2016a). The maximum F-measure achieved for the multimodal system for Hindi songs was 68.6% in Patra et al. (2016a). Indian music can be divided into two broad categories namely, “classical” and “popular” (Ujlambkar and Attar, 2012). Further, classical music tradition of India …

Content truncated to 500 words