随着信息检索技术的迅速发展，音乐检索将越来越重要。音频文件处理的未来发展方向将是尽快找出用户有兴趣的音频文件。本文在此基础上对传统的音频文件的分类方式及相应的查找方式提出革新:即根据用户已知的感兴趣的音频，由系统分析用户关注的音频特征，并由此为基础进行音频的搜索。本文将音频文件的分类属性由年代，歌手等附加信息扩充到音乐本身的音乐情感方面。另外本文还为音乐情感的分类提出了思路:即从语音情感分类及音乐赏析的角度，提出有效的音频特征.

With the fast developing of information retrieval technique, music retrieval will be more and more important.The future direction of the development of audio file processing will be identifing user interested audio files as soon as possible. Based on this idea,this thesis proposes an improvement over traditional way of audio files classification and related searching: according to the known audio files in which the user is interested, the system analysis the audio features relating to the interest, and searches the audio files according to such interest. The thesis proposes to extend to the way of analysising audio files from the traditional attributes, such as singer and era,to the music content and emotion. In addition, we propose an idea of music emotion classification: find the valid audio features from the aspect of speech emotion classification and music appreciation.