基于熵的音乐声纹检索算法的分析与实现-analysis and implementation of music voiceprint retrieval algorithm based on entropy.docxVIP

南京邮 南京邮电大学硕士研究生学位论文 Abstract I II Abstract With the development of network multimedia technology, there is more and more digital music on the internet. It has become a hot spot how to seek out target music by content-based music retrieval and not only by text limited. For a music record section, entropy-based retrieval method can find out the information about it quickly and accurately. With this method, you can search music without knowing its name. It has a great impact in the field of music arrangement and copyright protection. This thesis introduces the basic concepts and background about music retrieval. The key technique of music fingerprint extraction and search algorithm are researched. Based on them, some comparison experiments are also did. The main research works are as follows: First it analyzed the relevant theoretical knowledge of entropy-based retrieval method. It gives the basic framework and summarization about the search system. According to the point of view that one can’t know the distribution of music features in advance, the thesis gives the non-parametric music fingerprint estimation approach based on entropy. Under the usual distortion and noise conditions, the non-parametric method can describe the music distribution better and the fingerprint shows stronger robustness compared with parametric method. On the basis of the research and comparison among existing retrieval approaches, the thesis presents an improved matching search algorithm based on index. The experiment results show that this method gets higher success rate and improves the efficiency. According to the fingerprint extraction and retrieval algorithm researched in the paper，a simulating music retrieval system is designed and implemented. With the search segments dealt with software and in the actual conditions, the system achieved the better result. Keywords: Entropy, Non-parametric Estimation, Parzen Window Function, Index, Approximate Matching 南京邮 南京邮电大学硕士研究生学位论文 目录 PAGE