Music emotion classification for Turkish songs using lyrics

Abstract

Music has grown into an important part of people's daily lives. As we move further into the digital age in which a large collection of music is being created daily and becomes easily accessible renders people to spend more time on activities that involve music. Consequently, the form of music retrieval is changed from catalogue based searches to searches made based on emotion tags in order for easy and effective musical information access. In this study, it is aimed to generate a model for automatic recognition of the perceived emotion of songs with the help of their lyrics and machine learning algorithms. For this purpose, first 300 songs are selected and annotated by human taggers with respect to their perceived emotions. Thereafter, Unigram, Bigram and Trigram word features are extracted from song lyrics after performing text preprocessing where stemming of the Turkish words is an essential part. Then, term by document matrices are created where term frequencies and tf-idf scores are considered as representations for the indices. Five different classification algorithms are fed with these matrices in order to find the best combination that achieves the highest accuracy results where recall and precision values are used as comparison metrics. As a result, best accuracy results are obtained by using Multinomial Naïve Bayes classifier where Unigram features are used to create the term by document matrix. In this setting, Unigram features are stemmed by Zemberek Long stemming method, and the index representation is chosen as term frequency. For this combination, obtained recall and precision values are 43.7 and 46.9, respectively. Müzik insanlık tarihinde önemli bir yere sahiptir. Özellikle dijital çağda kişiler tarafından her gün yaratılan ve ulaşılan müzik koleksiyonlarının büyüklüğü ile müziğin önemi daha da artmış ve insanlar müzik içeren aktivitelere daha fazla zaman ayırmaya başlamışlardır. Bununla birlikte, müziğe bilgi geri getirim sürecini kolay ve etkin hale getirmek için yapılan katalog bazlı aramalar duygu tabanlı etiketlere göre aramalara dönüşmüştür. Bu araştırmada amacımız şarkı sözlerine göre bir şarkıdan algılanan duygunun otomatik olarak çıkarıldığı bir model geliştirmektir. Model metin bazlı sınıflandırma için kullanılan makina öğrenmesi algoritmaları ile oluşturulmuştur. Bu amaçla araştırmada 300 şarkı seçilmiş ve bu şarkılar kişiler tarafından hissedilen duygularına göre etiketlenmiştir. Devamında metin ön analizi ile şarkı sözleri Türkçe köklerine ayrıştırılarak Unigram, Bigram ve Trigram kelime özellikleri çıkartılmıştır. Ardından endeksleri terim sıklığı ve tf-idf değerleri olan doküman bazında terim matrisleri yaratılmıştır. Bu matris değerleri 5 farklı sınıflandırma algoritmasına girdi olarak verilerek en yüksek doğruluk sonuçları, hatırlama ve kesinlik metrikleri üzerinden araştırılmıştır. Araştırmanın sonucunda en yüksek kesinlik değeri Zemberek Uzun Kök Ayıştırma Metodu ile Unigram kelime özelliklerine göre ayrıştırılmış ve endeksi terim sıklığına göre belirlenmiş terim bazlı doküman m...

The aesthetic emotional expression of piano music art in the background of Internet of things

“…In this paper, it can be clearly seen from the Figure that the two types of emotions, HVLA emotion and HVHA emotion, have a clear degree of distinction. From the beginning to the end, there is no intersection, and the average node degree of HVLA is significantly higher than that of HVHA ( Durahim et al, 2018 ), that is, under high pleasure, there is a clear distinction between the level of arousal. Similarly, LVHA emotion and LVLA emotion, the distinction between the two is also more obvious.…”

Turkish Sentiment Analysis for Open and Distance Education Systems

“…SVM, ANN, KNN, Naive Bayes, and LR have mostly used classifiers in supervised learning based category. In the Turkish language, the supervised learning based methods were used in many SA application areas such as politics (Kaya et al, 2012), marketing (Sternberg et al, 2018), (Karahoca et al, 2019)), and music (Durahim et al, 2018).…”

Sentiment Analysis for Distance Education Course Materials: A Machine Learning Approach

“…They created 15 different emoji groups and divided the tweets into these emoji classes. Durahim et al (2018) conducted music classification studies in 2017. Predefined categories such as music genres and moods were created, and 45 Turkish popular artists were selected and labeling was done for the classification in 2 of 3 people if consensus was reached.…”