[1]张 燕,唐振民,李燕萍,等.基于MFCC和HMM的音乐分类方法研究[J].南京师范大学学报(工程技术版),2008,08(04):112-114.
　Zhang Yan,Tang Zhenmin,et al.Research of Music Classification Based on MFCC Feature and HMM Model[J].Journal of Nanjing Normal University(Engineering and Technology),2008,08(04):112-114.
点击复制

基于MFCC和HMM的音乐分类方法研究
分享到：

南京师范大学学报（工程技术版）[ISSN:1006-6977/CN:61-1281/TN]

卷:
08卷
期数:
2008年04期
页码:
112-114
栏目:
出版日期:
2008-12-30

文章信息/Info

Title:
Research of Music Classification Based on MFCC Feature and HMM Model
作者:
张 燕1;  2;  唐振民2;  李燕萍2;  邹 益2
1. 金陵科技学院信息技术学院, 江苏南京210006; 2. 南京理工大学计算机学院, 江苏南京210094
Author(s):
Zhang Yan1; 2; Tang Zhenmin2; Li Yanping2; Zou Yi2
1.College of Information Jinling Institution of Technology,Nanjing 210006,China;2.College of Computer Science,Nanjing University of Science and Technology,Nanjing 210094,China
关键词:
M el倒谱系数;  音乐分类;  隐马尔可夫模型
Keywords:
M e l frequency cepstrum coe fficients;  mus ic c lassifica tion;  hiddenM arkovm ode l
分类号:
TN912.34
摘要:
采用基于Mel倒谱系数特征的隐马尔可夫模型对音乐进行分类.对音乐通过有监督的学习方式进行聚类,分类时将测试样本归入似然值最大的类别,对同一音频抽取若干样本,对样本识别结果采用投票法判定该音频的音乐类别,使分类的准确率得到进一步的提高.仿真实验对4种分类器在有干扰和无干扰的环境下的分类性能进行了比较,实验结果表明该方法具有更好的抗干扰能力和正确率.
Abstract:
In th is paper, we use hiddenM arkovM odel based onM e-l frequency cepstrum coe ffic ients to c lassify the mus ic. Classification d iv ides the test sam ples into ca tego ries acco rding to the largest like lihood va lue. W e draw seve ra l samp les o f the sam em us ic frequency, identify the results o f the sam ples us ing the voting m ethod, and thus determ ine the ca tego ry o f the audio to further improve c lassifica tion accuracy. W e m ake a s imu la tion exper iment to compare the perform ance o f four different c lassifica tions in the env ironm en ts o f disturbance and nod istabance. The resu lts show that HMM class ification has m ore advantages on perform ance and is less sens itive to d isturbance

参考文献/References:

[ 1] Foo te J. An overview o f audio informa tion re tr ieva l[ J]. M ultim edia System s, 1999, 7( 1): 2-10.
[ 2] Foo te J. Content-based re trieval o f music and audio [ J]. M ultim edia S torage and A rchiv ing System II, 1997, 32 ( 29):138-147.
[ 3] Li S Z. Content-based c lassifica tion and re trieval of audio using the nearest feature linem ethod[ J]. IEEE Trans on Speech Audio Processing, 2000, 8( 5): 619-625.
[ 4] Lu Guo jun, Tem plarH. A techn ique tow ards autom atic aud io c lassifica tion and retr ieva l[ C] / / Proceed ings of the 4 th International
Conference on S ignal Processing. Be ijing: IEEE Xplore, 1998: 1 142-1 145.
[ 5] 卢坚, 陈毅松, 孙正兴, 等. 基于隐马尔可夫模型的音频自动分类[ J]. 软件学报, 2002, 8( 13): 1 593-1 597.
Lu Jian, Chen Y isong, Sun Zhengx ing, et a .l Autom atic aud io class ification by using hiddenM a rkov model[ J]. Jou rnal of Sofewa re, 2002, 8( 13): 1 593-1 597. ( in Ch inese)

备注/Memo

备注/Memo:
基金项目: 江苏省教育技术研究“十一五”规划重点课题( 2007 - I- 4704 )资助项目.
通讯联系人: 张 燕, 副教授, 博士研究生, 研究方向: 模式识别和音频信号处理. E-ma il: zy@ jit. edu. cn
更新日期/Last Update: 2013-04-24