计算机音乐分类辨识研究

摘要互联网和海量存储的发展，使得多媒体信息日益丰富，音频应用如语音识别、基于内容的音乐检索等都在开展相关的研究，计算机音乐的智能分类辨识重要性也逐日显现。分类辨识的发展和成果将对音乐的检索、推荐、音乐存储管理等具有重要的实际意义。为了在人工标注尽可能少的前提下，提高自动辨识分类的速度和准度，必须合理地提取不同音乐种类中有代表性的特征，并通过适当的机器学习方法建立分类辨识模型。音乐具有多样性和模糊性的特点，且分类的标准不尽相同，但在分类结果上来说，分类的结果越接近人的听觉感受系统就越优秀。本文主要通过对音乐的不同特征进行提取，比较分析不同的特征所辨识区分的能力，以及最终通过对不同特征的选取和组合，用支持向量机作为学习算法，建立分类辨识系统，从而对听觉感受上不同的音乐进行分类辨识。 68516
毕业论文关键词分类辨识特征提取支持向量机
Title Research of Computer Music Classification
Abstract That the rapidly growing Internet and mass storage has brought a huge amount of multimedia. Audio applications, especially speech recognition, content-based music retrieval and so on, are all being carried out research. The importance of intelligent classification of computer music is also being noticed. Classification’s development and achievements will have a great impact on music search, music recommendation and storage management, and it does have its practical significance. To try our best to reduce labeling by human labor, we need to improve the speed and accuracy of automatic classification, which requires extracting various features that can represent the music segments and establish an appropriate machine learning algorithm to build the model. However, music has its persity and ambiguity, and the classification criteria vary. But just take classification results into account, the closer to the human auditory system, the more excellent the system is. This paper is mainly about extractions of different features, and then gives a comparative analysis on the ability of recognition of distinct features, and finally select and combine those who has better performance as the input data of SVM to build the classification system.
Keywords feature extraction music classification SVM

1 绪论 ·· 1

1.1 研究背景 1

1.2 研究现状 1

1.3 论文结构及内容概述 · 2

2 人耳对音乐的认知机理 4

2.1 听觉的感知机制 4

2.2 听觉特性 4

3 音乐分类的情感模型 · 6

3.1 Hevner 模型 ·· 6

3.2 Russell 模型 · 7

3.3 Thayer 模型 ·· 8

4 音乐文件格式 10

4.1 MIDI · 10

4.2 WAV/AU 10

5 音乐的辨识特征 ·· 13

5.1 预处理 13

5.2 特征表示 ·· 15

6 辨识分类方法 20

6.1 支持向量机 · 20

7 模型训练及辨识过程 25

7.1 SVM 分类器 · 25

7.2 样本数据集 · 25

7.3 特征的宏观意义 ·· 26 计算机音乐分类辨识研究:http://www.751com.cn/jisuanji/lunwen_77074.html