【作者】 张良；

【导师】 李锵；

【作者基本信息】 天津大学 ， 信号与信息处理， 2010， 硕士

【摘要】 随着网络技术和多媒体技术的迅速发展,各种多媒体信息呈几何级数增长,人们也更有机会接触到大量的多媒体内容。音频是多媒体信息中重要的一种,如何自动地对这些音频数据进行管理就成为一个突出的问题。特别对于身边种类繁多的音乐数据,人们需要快速高效的方法对其进行识别和检索,以便于快速找到需要的音乐信息。本文研究的重点是应用压缩感知算法基于音乐自身特征的识别与检索,属于基于内容的音乐识别范畴,利用音乐的节奏特点达到识别不同音乐曲目的目的,节省了人为输入音乐主观信息的繁琐劳动,提高了音乐录入数据库的效率,扩展了音乐检索的方式。本文首先讨论了音乐的基本特性,并围绕音乐识别进行展开,提出了能代表音乐特点的并适合压缩感知理论的节奏特征作为识别特征。其次重点讨论了压缩感知算法在处理多媒体数据上的优势,以及应用于是音乐识别领域的可行性。最后通过仿真实验验证应用该算法进行音乐识别的可行性,测试了不同情况下音乐的识别效果。更多还原

【Abstract】 With the rapid development of the network technology and the multimedia technology, various kinds of information grow geometrically, and people are also more exposed to the large amount of multimedia content. Audio is one of the most important multimedia. One of the prominent issues is how to automatically manage the content of such audio media. Especially for many types of music, it requires fast and efficient way to identify them in order to find the necessary audio information quickly. This dissertation focused on music identification based on compression sensing algorithm.The focus of this thesis, music identification based on compressed sensing algorithm, which belongs to the field of content-based music identification. It not only saves the manual input work of music’s subjective information but also expands the way of music search to identify different music by the rhythm features of music.This thesis first discussed the basic characteristics of music and music identification, and then proposed several rhythm features which can be used with the compressed sensing theory as music identification features. Second, focused on the compressed sensing algorithm in dealing with perceived advantages of multimedia data, and identify areas of the music used in the feasibility. Finally, simulation results verify the feasibility of music identification based on compressed sensing theory, and tested many the different situations of music recognition.更多还原