E-LEARNING TOOLS FOR ETHNOMUSICOLOGICAL STUDENTS BASED ON MUSIC INFORMATION RETRIEVAL TECHNIQUES

Abstract

The content analysis of audio data is a paradigm in which algorithms are created and deducted from the data, allowing machines to "understand" the content of the audio signals and to process it further. This paper emphasis on applying musical information retrieval methods in order to provide a set of tools that could aid students in Musicology and Ethnomusicology. The analysis is done in a similar way that the famous composer Bela Bartok conducted his analytic study on folk songs, but making this process automatic through machine learning. Aesthetic reasoning has been used throughout this process in the attempt to synchronize the extracted relevant computed musicology data to ethnomusicology theory. By applying aesthetic means that are computed by the machine, we can regard this process an "Automated Artificial Musical Aesthete". Also the input data is being processed by using internal metrics that are relevant to a particular musical genre. The internal algorithm clusters this data finding characteristics that could be particular to more then one genere, thus leading to the link with the Musicology and Ethnomusicolgy fields that are concerned with the broad study of music, emphasizing on on more than one dimension. This set of tools could provide to students insight on the interconnections between the musical genres like social or cultural implications that were not so obvious at first, or simply provide metrics for each genre that could be the start for future research. The authors reused already established open source methods and developed a micro-system in which these tools become modules of a broader system.