A semantic metadata extraction for cross-cultural music environments by using culture-based music samples and filtering functions

In the field of music retrieval, one of the important issues is how to extract semantic metadata for culture-based ethnic music rather than only focusing on Western classical music. In this paper, we propose a method to extract semantic metadata for a variety of music. The significant features of our method are (1) normalization of tuning parameters based on a music data set when extracting music structural factors, and (2) use of typical music samples coinciding with creation of a cultural music filter to generate a musical elementimpression transformation. Finally, weighted impression words such as sad, happy, dreamy are extracted according to musical properties of each culture. Experimental results demonstrate the feasibility of our method and indicate how challenging this issue is.