基于ARM的音频分类系统设计与实现
|
摘要
随着计算机技术、通讯技术、网络技术的不断发展,多媒体数据正逐渐成为信息处理领域中主要的信息媒体形式。其中,音频信息是多媒体领域中重要的信息媒体,音频分类技术也逐渐成为近几年来语音处理领域的一个关键技术。音频系统是嵌入式系统应用中的一个重要组成部分,广泛应用于智能家居、智能手机等多媒体系统中。随着集成电路快速发展,嵌入式系统的功能变得越来越强大,音频分类技术在嵌入式系统上的应用越来越广泛普及。
本文在对现有的音频分类技术及其应用进行研究的基础上,设计并实现了一个基于ARM的音频分类系统。首先,对音频信息的主要来源及音频的语义进行了分析,对音频分类的关键技术,尤其是基于SVM的音频分类技术进行了探讨。其次,提出了一种音频片段和音频帧结合的音频特征分析与提取方法,并从帧层次和段层次上研究了音频的区别性特征,基于MFCC方法构造了音频特征集合,基于SVM分类器设计实现了音频分类模块。最后,设计并实现了一个嵌入式的音频分类系统,使音频分类技术在ARM上得到了应用,实现了音频驱动、音频训练和音频分类等功能,实现效果进一步验证了本文所提出的音频分类方法的有效性和可用性。
With the development of computer technology, communication technology, and network technology, the content and form of information processing has been more and more various. Multimedia processing plays an important role in information processing area. Audio information is important in the field of multimedia information media,audio classification technology has become a key technology during the field of speech processing in recent years.the audio system become an important part of the embedded system applications widely used in smart home, public places, smart phones and other multimedia systems. With the rapid development of integrated circuits, the embedded system functions become more and more powerful, and audio classification becomes more widely popular in embedded systems.
In this paper, we designed and implemented an ARM-based audio classification system based on the study of technology and its application of the existing audio classification. First, we inspect the main source of audio information, analysi of the semantic content of the audio, discusse the key technologies of audio classification especially the SVM-based audio classification technology. Second, after studying the distinctive features of the audio from the frame level and segment level, we propose a feature analysis and extraction method through combining the audio clip and audio frame, using the MFCC method to construct a set of audio features, and study of the core issues of the SVM classifier in-depth, relize the audio classifier module based on SVM classification. Finally, this paper designs and implements an audio classification system to illustrate the application of the the audio classification technology on the ARM, including audio drivers, audio training, audio classification techniques to achieve further validate the effectiveness and availability of the audio feature classification.
本文在对现有的音频分类技术及其应用进行研究的基础上,设计并实现了一个基于ARM的音频分类系统。首先,对音频信息的主要来源及音频的语义进行了分析,对音频分类的关键技术,尤其是基于SVM的音频分类技术进行了探讨。其次,提出了一种音频片段和音频帧结合的音频特征分析与提取方法,并从帧层次和段层次上研究了音频的区别性特征,基于MFCC方法构造了音频特征集合,基于SVM分类器设计实现了音频分类模块。最后,设计并实现了一个嵌入式的音频分类系统,使音频分类技术在ARM上得到了应用,实现了音频驱动、音频训练和音频分类等功能,实现效果进一步验证了本文所提出的音频分类方法的有效性和可用性。
With the development of computer technology, communication technology, and network technology, the content and form of information processing has been more and more various. Multimedia processing plays an important role in information processing area. Audio information is important in the field of multimedia information media,audio classification technology has become a key technology during the field of speech processing in recent years.the audio system become an important part of the embedded system applications widely used in smart home, public places, smart phones and other multimedia systems. With the rapid development of integrated circuits, the embedded system functions become more and more powerful, and audio classification becomes more widely popular in embedded systems.
In this paper, we designed and implemented an ARM-based audio classification system based on the study of technology and its application of the existing audio classification. First, we inspect the main source of audio information, analysi of the semantic content of the audio, discusse the key technologies of audio classification especially the SVM-based audio classification technology. Second, after studying the distinctive features of the audio from the frame level and segment level, we propose a feature analysis and extraction method through combining the audio clip and audio frame, using the MFCC method to construct a set of audio features, and study of the core issues of the SVM classifier in-depth, relize the audio classifier module based on SVM classification. Finally, this paper designs and implements an audio classification system to illustrate the application of the the audio classification technology on the ARM, including audio drivers, audio training, audio classification techniques to achieve further validate the effectiveness and availability of the audio feature classification.
引文
[1]马华东.多媒体技术原理和应用.清华大学出版社,2008
[2]黄王.基于支持向量机音频分类技术研究与应用.北京邮电大学,2010
[3]王超.基于小波和隐马尔可夫模型的音频分类.西北工业大学,2007
[4]纪正飚.基于ARM技术的语音识别研究.河海大学,2007
[5]Huang X, Acero A, Hon H W.Spoken Language Processing:A Guide to Theory, Algorithem and System Development.2001
[6]Wilpin J, Rabiner L R, Lee C H, eal. Automatic recognition of keywords in unconstrained speech using hidden Markov models. IEEE Transactions on Applied Super-conductivity.1992
[7]S.Pruzansky.Pattem-matching procedure for automatic talke recognition. J Acoust. Sec. Amer.1963
[8]罗建荣.多类统计模式识别模型及应用研究.重庆大学,2009
[9]王荣燕.复杂音频分类中的关键问题研究.北京邮电大学,2011
[10]张永.基于模糊支持向量机的多类分类算法研究.大连理工大学,2007
[11]李忠伟.支持向量机学习算法研究.哈尔滨工程大学,2005
[12]张凤羽.基于内容的分布式音频检索技术研究与实现.北京邮电大学,2012
[13]周兴恩.基于GMM和SVM的音频分类算法.大连理工大学,2008
[14]白亮.音频分类与分割技术研究.国防科学技术大学,2004
[15]贺玲玲,周元.基于改进MFCC的异常声音识别算法.重庆工商大学学报,2012年2月
[16]D. R4J REDDY, Speech Recognition by Machine:A Review [J], PROCEEDINGS OF 1976:505-506
[17]丁子春.基于支持向量机的多领域自适应分类方法研究.武汉理工大学,2009
[18]刘乔辉.基于神经网络的5种音频分类.中国西部科技,2008年3月
[19]俞玉莲,郭世杰.音频分类中的特征分析.信息技术,2009年6期
[20]高翔.超球支持向量机在语音识别中的应用研究.太原理工大学,2011
[21]王荣燕.复杂音频分类中的关键问题研究.北京邮电大学,2011
[22]程小梅.基于图像的昆虫识别研究与设计.西北大学,2008
[23]常梦星.多媒体数据库中基于内容的音频分类综述.电脑编程技巧与维护,2010年7月
[24]潘越群.时间表达式识别与归一化研究.哈尔滨工业大学,2008
[25]邢慧强.SVM用于块划分特征提取的图像分类与检索.中国海洋大学,2006
[26]陈少英.支持向量机算法在邮件过滤中的应用.福建电脑,2007
[27]张晓飞.基于支持向量机的图像信息分类与检索.沈阳大学,2007
[28]张巍.基于ARM9的嵌入式汽车导航与防盗系统的设计.大连海事大学,2009
[29]张杰.基于Linux的ARM9嵌入式系统设计.南京理工大学,2007
[30]张喜龙.基于ARM9的嵌入式Linux应用开发平台的分析与实现.大连海事大学,2008
[31]郭秋平.基于ARM系统的Linux平台移植研究.浙江大学,2006
[32]周庆华.基于嵌入式Linux说话人识别系统的研究.西华大学,2009
[33]Huang X, Acero A, Hon H W. Spoken Language Processing:A Guide to Theory, Algorithem and System Development.2001
[34]张永.基于模糊支持向量机的多类分类算法研究.大连理工大学,2007
[35]Inn H. Witten, Eibe Frank数据挖掘：实用机器学习技术.机械工业出版社,2009
[36]罗建荣.多类统计模式识别模型及应用研究.重庆大学,2009
[37]霍燃,高丽萍,陈庆奎.嵌入式Linux系统下基于UDA1341芯片的音频驱动程序设计.计算机应用与软件, Vol 29 No.4 Apr.2012
[2]黄王.基于支持向量机音频分类技术研究与应用.北京邮电大学,2010
[3]王超.基于小波和隐马尔可夫模型的音频分类.西北工业大学,2007
[4]纪正飚.基于ARM技术的语音识别研究.河海大学,2007
[5]Huang X, Acero A, Hon H W.Spoken Language Processing:A Guide to Theory, Algorithem and System Development.2001
[6]Wilpin J, Rabiner L R, Lee C H, eal. Automatic recognition of keywords in unconstrained speech using hidden Markov models. IEEE Transactions on Applied Super-conductivity.1992
[7]S.Pruzansky.Pattem-matching procedure for automatic talke recognition. J Acoust. Sec. Amer.1963
[8]罗建荣.多类统计模式识别模型及应用研究.重庆大学,2009
[9]王荣燕.复杂音频分类中的关键问题研究.北京邮电大学,2011
[10]张永.基于模糊支持向量机的多类分类算法研究.大连理工大学,2007
[11]李忠伟.支持向量机学习算法研究.哈尔滨工程大学,2005
[12]张凤羽.基于内容的分布式音频检索技术研究与实现.北京邮电大学,2012
[13]周兴恩.基于GMM和SVM的音频分类算法.大连理工大学,2008
[14]白亮.音频分类与分割技术研究.国防科学技术大学,2004
[15]贺玲玲,周元.基于改进MFCC的异常声音识别算法.重庆工商大学学报,2012年2月
[16]D. R4J REDDY, Speech Recognition by Machine:A Review [J], PROCEEDINGS OF 1976:505-506
[17]丁子春.基于支持向量机的多领域自适应分类方法研究.武汉理工大学,2009
[18]刘乔辉.基于神经网络的5种音频分类.中国西部科技,2008年3月
[19]俞玉莲,郭世杰.音频分类中的特征分析.信息技术,2009年6期
[20]高翔.超球支持向量机在语音识别中的应用研究.太原理工大学,2011
[21]王荣燕.复杂音频分类中的关键问题研究.北京邮电大学,2011
[22]程小梅.基于图像的昆虫识别研究与设计.西北大学,2008
[23]常梦星.多媒体数据库中基于内容的音频分类综述.电脑编程技巧与维护,2010年7月
[24]潘越群.时间表达式识别与归一化研究.哈尔滨工业大学,2008
[25]邢慧强.SVM用于块划分特征提取的图像分类与检索.中国海洋大学,2006
[26]陈少英.支持向量机算法在邮件过滤中的应用.福建电脑,2007
[27]张晓飞.基于支持向量机的图像信息分类与检索.沈阳大学,2007
[28]张巍.基于ARM9的嵌入式汽车导航与防盗系统的设计.大连海事大学,2009
[29]张杰.基于Linux的ARM9嵌入式系统设计.南京理工大学,2007
[30]张喜龙.基于ARM9的嵌入式Linux应用开发平台的分析与实现.大连海事大学,2008
[31]郭秋平.基于ARM系统的Linux平台移植研究.浙江大学,2006
[32]周庆华.基于嵌入式Linux说话人识别系统的研究.西华大学,2009
[33]Huang X, Acero A, Hon H W. Spoken Language Processing:A Guide to Theory, Algorithem and System Development.2001
[34]张永.基于模糊支持向量机的多类分类算法研究.大连理工大学,2007
[35]Inn H. Witten, Eibe Frank数据挖掘：实用机器学习技术.机械工业出版社,2009
[36]罗建荣.多类统计模式识别模型及应用研究.重庆大学,2009
[37]霍燃,高丽萍,陈庆奎.嵌入式Linux系统下基于UDA1341芯片的音频驱动程序设计.计算机应用与软件, Vol 29 No.4 Apr.2012