基于ARM的音频分类系统设计与实现

发布时间：2018-04-24 04:33

本文选题：音频分类 + 特征提取　；参考：《北京邮电大学》2013年硕士论文

【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。
[Abstract]:With the development of computer technology, communication technology and network technology, multimedia data is becoming the main information media in the field of information processing. Among them, audio information is an important information medium in the field of multimedia, and audio classification technology has gradually become a key technology in the field of speech processing in recent years. Audio system is an important part of embedded system application, widely used in smart home, smart phone and other multimedia systems. With the rapid development of integrated circuits, the functions of embedded systems become more and more powerful, and audio classification technology is more and more widely used in embedded systems. Based on the research of the existing audio classification technology and its application, an audio classification system based on ARM is designed and implemented in this paper. Firstly, the main sources of audio information and the semantics of audio are analyzed, and the key technologies of audio classification, especially the audio classification based on SVM, are discussed. Secondly, an audio feature analysis and extraction method based on the combination of audio segment and audio frame is proposed, and the distinguishing features of audio are studied from frame level and segment level, and the audio feature set is constructed based on MFCC method. Audio classification module based on SVM classifier is designed and implemented. Finally, an embedded audio classification system is designed and implemented. The audio classification technology is applied on ARM, and the functions of audio driver, audio training and audio classification are realized. The effectiveness and availability of the proposed audio classification method are further verified.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP368.1;TN912.3

【参考文献】