{$cfg_webname}
主页 > 电子信息 > 电子 >

基于MATLAB的语音增强的研究-谱减算法与子空间算法的比较

来源:wenku163.com  资料编号:WK16313915 资料等级:★★★★★ %E8%B5%84%E6%96%99%E7%BC%96%E5%8F%B7%EF%BC%9AWK16313915
资料介绍

基于MATLAB的语音增强的研究-谱减算法与子空间算法的比较(论文12000字)
摘要:语音作为一种典型的非平稳随机信号,是人类进行信息交流的最自然、最有效、最方便的方法。它在人类文明和社会进步中发挥着着重要作用。语音的传输、识别和增强等是数字通信网络中最重要的组成部分之一,也是最基本的一种。语音技术的实际应用的增多也导致了处理各种语音系统时必须正视问题,即要进一步提高性能。语音增强就是解决提高性能这一问题的关键技术之一。当前语音增强分为两大类方法。一是时域方法(子空间算法);二是频域方法(如谱减法等)。两类方法都有各自的优点和缺点:子空间算法提供了一种机制来控制语音信号的失真程度和残留噪声的最小化,但计算复杂度大。另一方面,谱减法的计算是小的,但没有理论机制的语音信号失真和残余噪声的控制。 在本文中,我们主要研究谱减算法和子空间算法的理论知识以及它们在语音增强中的应用,目的是比较这两种算法,以提高语音质量,减少语音失真,提高整体质量的可读方面。
关键字:语音增强;谱减法;子空间算法;语音质量

Research on Speech Enhancement Based on MATLAB
----- Comparison of Spectral Subtraction Algorithm and Subspace Algorithm
Abstract: speech, as a typical non-stationary random signal, is the most convenient method for human to communicate with each other. It plays an important role in human civilization and social progress. Voice interaction has become a necessary means of human-computer interaction, voice transmission, digital recognition, synthesis and enhancement is one of the most important part of the digital communication network, is one of the most basic. The deep research and practical application of speech technology also lead to the problem that all kinds of speech processing systems must face up, that is, to further improve the performance. Speech enhancement is one of the key technologies to solve this problem. Currently, it is divided into two broad categories. One is time domain method (subspace algorithm), and two is frequency domain method (such as spectral subtraction). Subspace algorithm provides a mechanism to control the distortion of speech signals and minimize the residual noise, but the computational complexity is large. On the other hand, spectral subtraction calculations are small, but there is no theoretical mechanism for speech signal distortion and residual noise control. In this paper, we mainly study the theory of spectral subtraction algorithm and subspace algorithm and their application in speech enhancement, is to compare the two algorithms to improve the speech quality, reducing speech distortion, improve the overall quality of the readability.
Key words:speech enhancement; spectral subtraction; subspace algorithm; speech quality
 

基于MATLAB的语音增强的研究-谱减算法与子空间算法的比较


目录
引言    1
一、语音增强的基础知识    1
1.1语音信号    1
1.2噪声    1
1.3语音增强算法的分类    2
1.4语音信号预处理    2
1.4.1加窗分帧    2
1.4.2短时傅里叶变换    2
二、谱减算法    3
2.1 基本谱减算法    3
2.2 使用过减技术    4
2.3多带谱减算法    5
2.4 采用自适应增益平均和低延时卷积的谱减计算计算更新噪声谱    7
三、子空间算法    9
3.1 基于EVD的方法:色噪声    9
3.1.1 预白化方法    9
3.1.2 内嵌预白化的子空间方法    11
3.2 基于感知的子空间算法    12
3.2.1傅里叶变换域与特征域的关系    12
3.2.2加入心理声学模型约束    13
四、MATLAB简介与仿真    14
4.1 MATLAB简介    14
4.1.1软件特点    14
4.1.2 MATLAB语言的优点    15
4.2  MATLAB的仿真    15
五、两种算法客观音质测度的比较    18
5.1 客观音质测度    18
5.1.1 分段信噪比    18
5.1.2 基于LPC的谱距离测度    19
5.1.3加权谱斜率(WSS)距离测度    20
5.1.4 感知语音质量评估(PESQ)测度    20
5.2 音质可观测度结果    22
六、结论    24
七、参考文献    24
致谢    26

推荐资料