Music Retrieval Based On Music Topic

With the rapid development of Internet multimedia technology and Internet,a large number of digital music has been expanded on the Internet, allowing users to have a lot of options to receive music, but it also brings new challenges for Music Information Retrieval(MIR) system. Text-based retrieval is the traditional technology in the search engine, which depends on the musical attributes only to indirectly receive music. It is clearly that the music retrieval algorithm like this has been unable to meet the growing intelligent, personalized search needs, So content-based music retrieval technology came into being. A query by humming algorithm is the mainstream way in the study of content-based music retrieval technique, but this approach is so specialized that it is used widely by the general users.With the development of Web 2.0 and collaboratively tagging function,users can annotate and categorize their own interests and thoughts about the resources possible. It makes the social tags which has a relatively high semantic information to be the important resources in the music sharing system. If we simply complete simple music retrieval based on music tags, which will not only ignore the semantic relevance between tag information, but also ignore the user’s true retrieval intention. Considering the characteristics of social tags and its influence in Music Information Retrieval(MIR) system, this paper has put forward music retrieval based on music topic.Music retrieval based on music topic mainly used MT- LDA model to cluster music tags and train topic model parameters, by which the search engine could analysis the input of user to catch the retrieval topic to realize the user’s personal retrieval demand, so as to improve the efficiency and performance of MIR system. In the paper, a result had been found that music retrieval based on music topic can understand the user’s requirements and intention to retrieval more content which could meet the user,s requirement. What ’s more,it performed better than keywords retrieval to improve performance of the MIR system especially when retrieval information had no attribute data.