Sarkı Sözü ve Ses Niteliklerini Kullanarak Türkçe Müzik Türü Sınıflandırması

Müzik Bilgi Getirimi (MIR) son yıllarda popüler bir araştırma alanı olmuştur. Bu baglamda, araştırmacılar müzik türü, sevilen şarkıların tespiti ve otomatik çalma listesi oluşturma gibi önemli problemlere çözüm üretmek için müzik bilgi sistemleri geliştirmişlerdir. Önceki çalışmalarda üst-veri bilgisi, şarkı sözleri ya da müzigin melodik içerigi nitelik kaynagı olarak kullanılmıştır. Ancak, şarkı sözleri genellikle MIR sistemlerinde ˘ kullanılmamış ve özellikle Türkçe için bu alanda yapılan çalışma sayısı yetersiz kalmıştır. Bu çalışmada, ilk olarak, her bir şarkıya ait ses dosyası eklenerek daha önce oluşturdugu- ˘ muz Türkçe şarkı sözlerinden oluşan Türkçe MIR (TMIR) veri kümesi genişletilmiştir. ˙Ikinci olarak, ses ve metinsel niteliklerin birlikte ve ayrı kullanıldıklarında Müzik Türü Sınıflandırması (MGC) üzerindeki etkisi incelenmiştir. Metinsel nitelikler word2vec ve kelime torbası gibi nitelik çıkarım modelleri ile şarkı sözlerinden çıkarılmıştır. Deneyler Destek Vektör Makinesi (SVM) algoritması ile gerçekleştirilmiş ve nitelik seçimi ile farklı nitelik gruplarının MGC üzerindeki etkisi incelenmiştir. şarkı sözü tabanlı MGC bir metin sınıflandırma işlemi olarak ele alınmış, ayrıca terim agırlıklandırma yönteminin etkisi incelenmiştir. Deneysel sonuçlar ses niteliklerinin yanı sıra özellikle denetimli bir agırlık- landırma yöntemi kullanıldıgında metinsel niteliklerin de MGC için etkili olabilecegini göstermiştir. Metinsel nitelikler ses nitelikleri ile birlikte kullanılarak en yüksek %99,12 oranında başarı elde edilmiştir.

Turkish Music Genre Classification using Audio and Lyrics Features

Music Information Retrieval (MIR) has become a popular research area in recent years. In this context, researchers have developed music information systems to find solutions for such major problems as automatic playlist creation, hit song detection, and music genre or mood classification. Meta-data information, lyrics, or melodic content of music are used as feature resource in previous works. However, lyrics do not often used in MIR systems and the number of works in this field is not enough especially for Turkish. In this paper, firstly, we have extended our previously created Turkish MIR (TMIR) dataset, which comprises of Turkish lyrics, by including the audio file of each song. Secondly, we have investigated the effect of using audio and textual features together or separately on automatic Music Genre Classification (MGC). We have extracted textual features from lyrics using different feature extraction models such as word2vec and traditional Bag of Words. We have conducted our experiments on Support Vector Machine (SVM) algorithm and analysed the impact of feature selection and different feature groups on MGC. We have considered lyrics based MGC as a text classification task and also investigated the effect of term weighting method. Experimental results show that textual features can also be effective as well as audio features for Turkish MGC, especially when a supervised term weighting method is employed. We have achieved the highest success rate as 99,12% by using both audio and textual features together.

PDF

___

McKay, C., Burgoyne, J. A., Hockman, J., Smith, J. B., Vigliensoni, G., Fujinaga, I. 2010. Evaluating the Genre Classification Performance of Lyrical Features Relative to Audio, Symbolic and Cultural Features. ISMIR, 9-13 August, Utrecht, 213-218.
Kohavi, R. 1995. A study of cross-validation and bootstrap for accuracy estimation and model selection. In Ijcai, 14(1995), 1137-1145.
Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., Witten, I. H. 2009. The WEKA data mining software: an update. ACM SIGKDD explorations newsletter, 11(2009), 10-18.
Kotsiantis, S. B., Zaharakis, I., Pintelas, P. 2007. Supervised machine learning: A review of classification techniques. In Proceedings of the 2007 conference on Emerging Artificial Intelligence Applications in Computer Engineering, 3-24.
Joachims, T. 1998. Text categorization with support vector machines: Learning with many relevant features. In European conference on machine learning, 21-23 April, Chemnitz, 137-142.
Hall, M. A. 2000. Correlation-based feature selection for discrete and numeric class machine learning. In Proceedings of the Seventeenth International Conference on Machine Learning, 29 June-2 July, Stanford, 359–366.
Lan, M., Tan, C. L., Su, J., Lu, Y. 2009. Supervised and traditional term weighting methods for automatic text categorization. IEEE transactions on pattern analysis and machine intelligence, 31(2009), 721-735.
Kansheng, S. H. I., Jie, H. E., Liu, H. T., Zhang, N. T., Song, W. T. 2011. Efficient text classification method based on improved term reduction and term weighting. The Journal of China Universities of Posts and Telecommunications, 18(2011), 131-135.
Salton, G., Wong, A., Yang, C. S. 1975. A vector space model for automatic indexing. Communications of the ACM, 18(1975), 613-620.
Yuan, Y., He, L., Peng, L., Huang, Z. 2014. A new study based on word2vec and cluster for document categorization. Journal of Computational Information Systems, 10(2014), 9301-9308.
Mikolov, T., Sutskever, I., Chen, K., Corrado, G. S., Dean, J. 2013. Distributed representations of words and phrases and their compositionality. In Advances in neural information processing systems, 5-10 December, Lake Tahoe, 3111-3119.
Mikolov, T., Chen, K., Corrado, G., Dean, J. 2013. Efficient estimation of word representations in vector space. In 2013 International Conference on Learning Representations (ICLR), 2-4 May, Scottsdale.
Mayer, R., Neumayer, R., Rauber, A. 2008. Combination of audio and lyrics features for genre classification in digital audio collections. In Proceedings of the 16th ACM international conference on Multimedia, 26-31 October, Vancouver, 159-168
Stamatatos, E., Fakotakis, N., Kokkinakis, G. 2000. Automatic text categorization in terms of genre and author. Computational linguistics, 26(2000), 471-495.
Lodhi, H., Saunders, C., Shawe-Taylor, J., Cristianini, N., Watkins, C. 2002. Text classification using string kernels. Journal of Machine Learning Research, 2(2002), 419-444.
Kanaris, I., Kanaris, K., Houvardas, I., Stamatatos, E. 2006. Words Vs Characters N-Grams for AntiSpam Filtering. International Journal on Artificial Intelligence Tools, world Scientific, X(2006), 1-20.
Lewis, D. D. 1992. An evaluation of phrasal and clustered representations on a text categorization task. In Proceedings of the 15th annual international ACM SIGIR conference on Research and development in information retrieval, 21-24 June, Copenhagen, 37-50.
Joachims, T. 1997. A Probabilistic Analysis of the Rocchio Algorithm with TFIDF for Text Categorization. Proceedings of the Fourteenth International Conference on Machine Learning (ICML ’97 ), 8-12 July, 143-151.
Öztürk, M. B., Can, B. 2016. Clustering word roots syntactically. In 2016 24th Signal Processing and Communication Application Conference (SIU), 16-19 May, Zonguldak, 1461-1464
Çoban, Ö., Özyer, G. T., 2016. Sentiment classification for Turkish Twitter feeds using LDA. In 2016 24th Signal Processing and Communication Application Conference (SIU), 16-19 May, Zonguldak, 129-132.
Akın, A. A., Akın, M. D. 2007. Zemberek, an open source NLP framework for Turkic languages. Structure, 10(2007), 1-5.
Lin, Y., Lei, H., Wu, J., Li, X. 2015. An Empirical Study on Sentiment Classification of Chinese Review using Word Embedding. PACLIC 29, 30 October- 1 November, Shanghai, 258-266.
Hu, X., Downie, J. S., Ehmann, A. F. 2009 Lyric text mining in music mood classification. American music, 183(2009), 2-209.
McEnnis, D., McKay, C., Fujinaga, I., Depalle, P. 2006. jAudio: Additions and Improvements. In ISMIR, 8-12 October, Victoria, 385-386.
McKay, C., Fujinaga, I., Depalle, P. 2005. jAudio: A feature extraction library. In Proceedings of the International Conference on Music Information Retrieval, 11-15 September, London, 600-3.
McKay, C., Fujinaga, I. 2008. Combining Features Extracted from Audio, Symbolic and Cultural Sources. In ISMIR, 14-18 September, Philadelphia, 597-602.
Cataltepe, Z., Yaslan, Y., Sonmez, A. 2007. Music genre classification using MIDI and audio features. EURASIP Journal on Advances in Signal Processing, 2007(2007), 1-8.
Dhanaraj, R., Logan, B. 2005. Automatic Prediction of Hit Songs. In ISMIR, 11-15 September, London, 488-491.
Yaslan, Y., Cataltepe, Z. 2006. Audio music genre classification using different classifiers and feature selection me thods. In 18th International Conference on Pattern Recognition (ICPR’06), 20-24 August, Hong Kong, 573-576
Ogul, H., Kırmacı, B. 2016. Lyrics Mining for Music ˘ Meta-Data Estimation. In IFIP International Conference on Artificial Intelligence Applications and Innovations. Thessaloniki, 16-18 September, 528-539.
Van Zaanen, M., Kanters, P. 2010. Automatic Mood Classification Using TF* IDF Based on Lyrics. In ISMIR, 9-13 August, Utrecht, 75-80.
Mayer, R., Rauber, A. 2010. Multimodal Aspects of Music Retrieval: Audio, Song Lyrics–and Beyond?. In Advances in Music Information Retrieval, 333-363
Yang, D., Lee, W. S. 2009. Music emotion identifi- cation from lyrics. In ISM’09, 14-16 December, San Diego, 624-629
Zheng, F., Zhang, G., Song, Z. 2001. Comparison of different implementations of MFCC. Journal of Computer Science and Technology, 16(2001), 582- 589.
Mayer, R., Neumayer, R., Rauber, A. 2008. Rhyme and Style Features for Musical Genre Classification by Song Lyrics. In ISMIR, 14-18 September, Philadelphia, 337-342.
Kırmacı, B., Ogul, H. 2015. Author recognition from lyrics. Signal Processing and Communications Applications Conference (SIU), 16-19 May, Malatya, 2489- 2492.
Kızrak, M. A., Bayram, K. S., Bolat, B. 2014. Classification of Classic Turkish Music Makams. In Innovations in Intelligent Systems and Applications (INISTA) Proceedings, 23-25 June, Alberobello, 394-397
Alpkoçak, A., Gedik, A. C. 2006. Classification of Turkish songs according to makams by using n grams. In Proceedings of the 15. Turkish Symposium on Artificial Intelligence and Neural Networks (TAINN), Mugla
Holzapfel, A., Stylianou, Y. 2009. Rhythmic Similarity in Traditional Turkish Music. In ISMIR, 26-30 October, Kobe, 99-104.
Çoban, Ö., Özyer, G. T., 2016. Music genre classifi- cation from Turkish lyrics. In 2016 24th Signal Processing and Communication Application Conference (SIU), 16-19 May, Zonguldak, 101-104
Ying, T. C., Doraisamy, S., Abdullah, L. N. 2012. Genre and mood classification using lyric features. In Information Retrieval & Knowledge Management (CAMP), 13-15 March, Kuala Lumpur, 260-263.
Hu, X., Downie, J. S. 2010. Improving mood classification in music digital libraries by combining lyrics and audio. In Proceedings of the 10th annual joint conference on Digital libraries, 21-25 June, Gold Coast, QLD, 159-168
Sordo, M. 2012. Semantic annotation of music collections: A computational approach. Universitat Pompeu Fabra, Department of Information and Communication Technologies, Doctoral dissertation, 18p, Barcelona.