期刊名称：International Journal of Computer Trends and Technology
电子版ISSN：2231-2803
出版年度：2016
卷号：36
期号：1
页码：17-21
DOI：10.14445/22312803/IJCTT-V36P104
出版社：Seventh Sense Research Group
摘要：One of the problems to solve in Music Information Retrieval (MIR) is the modelization of music style. The system could be trained to identify the main features that would characterize music genres or style so as to look for that kind of music over large musical corpus. So in this paper multimodal approach, pattern recognition approach and coupdating approach is been studied for identifying the style from different genre of the music. Considering the intuitive feelings of similarity from the listeners perspective, the focus on features that are computed using similarity metrics for melodies, harmonies, and audio signals for style identification. A multimodal approach mostly considered support vector machine as a binary classifier to determine if two songs or music played by the same artist given their similarity metrics in the three aspects and also discussed the experimental methodologies of the two different approaches.
关键词：Gaussian mixture models; melodiccontour; music similarity; n-grams; style.
