Research On Several Problems Of Computing Music

The development of culture computing has produced a great impact on data mining, asa representative of culture, music computing has increasingly become a research hotspot.Before MusicXML occurred, the format of music files on Internet had existed in the formof audio, and there has no uniform standard between different equipment and network usedfor exchanging music files, so researching music files has been a problem. MusicXMLbases on mature technology of XML, providing a standard data storage scheme for musicfiles, and at the same time, it provides a solution for mining music in computing.Research of computing music cannot do without the study of feature extraction. In thispaper, we calculated music files based on MusicXML, improving a portion of its structure,and then designed an algorithm according to certain rules to extract music genes. Then weapplied music genes to classification and similarity matching. Our main research includesthe status of computing music, the extraction rules and algorithm, the classification andsimilarity matching using music genes.This article first invested the domestic and foreign research of music mining in thefield of computing music, discussing their method and data model, and then put forward thesignificance for this topic according to present research’s shortcomings. In the third chapter,this paper improved a portion of the structure of XML, remaining more complete melodyinformation when extracting music feature. After obtaining music genes, they were appliedto the classification of era feature. During the classification, in order to take the structure and content of music genes into consideration, discussions for various classifiers were ma-de and the support vector machine was selected finally. The experimental part discussed theaccuracies under different circumstances, results show that music genes can reflectthe era attribute of music in a certain extent. In the end, this article attemptedto discuss similarity comparison using music genes, taking both music genes and theirXML structure into account. Experiment compared the matching count using similari-ty, three kinds of different methods showed that, the music genes and their structure per-formed better. Although its matching number was small, their matching results were moreaccurate.