Classification Of Sinhala Songs Based On Emotions
2019 19TH INTERNATIONAL CONFERENCE ON ADVANCES IN ICT FOR EMERGING REGIONS (ICTER - 2019)(2019)
摘要
Classification of Sinhala Songs has received less attention from researchers in the Sri Lankan context. The purpose of the current study is to shed light on emotion based classification based on Music Emotion Recognition (MER) which is a sub domain of Music Information Retrieval (MIR). In achieving this the authors followed three steps namely; selection of an appropriate emotion model, extraction and selection of low level music feature and training of the classifier utilizing supervised machine learning algorithms. 18 experiments were conducted to analyze the level of accuracy in each feature selection algorithm in combination with the supervised machine learning algorithms. Results of the current study suggest that the combination of ReliefF based feature selection algorithm and Random Forest supervised machine learning algorithm yields highest accuracy classification of Sinhala songs based emotions. The highest accuracy level of 91.98% in the current study is identified as the highest result achieved within the domain of Sinhala songs classification according to literature review conducted.
更多查看译文
关键词
Music Information Retrieval, Music Emotion Recognition, Music Feature Extraction, Music Feature Selection, Supervised Machine Learning
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要