共查询到20条相似文献，搜索用时 906 毫秒
1.
针对语音信号降噪和提取算法存在降噪效果不佳等问题,已经无法满足现在社会的需求。本文提出一种基于改进EMD算法和Hilbert算法混合的语音信号降噪和提取方法,首先针对EMD算法在语音信号降噪可能破坏噪声特性的问题,利用全局阈值选择去噪的方法对其进行改进,然后引入Hil-bert算法将其与EMD算法融合,对语音信号进行降噪和提取。利用仿真实验可以发现:将EMD算法和Hilbert算法进行混合优化,和传统的小波变换语音信号降噪算法,不容易出现差错并且具有更好的降噪性能。 相似文献
2.
3.
4.
5.
独立分量分析是近年发展起来的一种高效的信号分离方法,主要对观测的混合信号进行分离或提取各个源信号。简要介绍了ICA的概念、基本原理以及FastICA算法,通过实际语音信号的仿真,证明了用FastICA算法分离语音信号可以取得较好的结果。 相似文献
6.
语音信号技术广泛应用于语音通信、语音识别等领域,在语音信号处理过程中,降噪与提高可懂度是极为关键的,处理方法有多种,通过综合梳理,对谱减法和自适应噪声抵消法的分析,发现最大限度地消除谱减法中的"音乐噪声"成为需要突破的研究域。在此尝试利用MATLAB以图像形式呈现谱减法增强语音信号效果的比较。发现语音信号处理算法在定义域内,仍有不可避免的误差。得出避免误差算法的加强与从强噪声中提取语音信号,既是语音信号处理过程中所"遭遇"的重要问题,又是未来理论与实践应用研究的发展方向。 相似文献
7.
在语音识别系统中,往往需要对输入的语音信息进行数据预处理操作,删除冗余的、不相关的特征值.针对传统应用于语音系统中特征选择算法中出现的效率低、错误率高的缺点,本文提出了基于信息增益的特征选择算法.该算法通过信息增益评价指标对属性进行排名及评价,选择最优的特征属性并删除无用的属性.通过大量的对比实验结果表明,本文提出的算法可以高效地完成特征选择语音数据预处理,并且提出的新算法与传统的特征选择算法选择出的特征属性应用在语音识别算法后能够更准确地识别和判断语音信息. 相似文献
8.
分析了信号子空间方法可以提高信噪比的原因,讨论了信号子空间方法在语音增强中的应用及信号子空间分解的常用方法.同时,将将单个麦克风信号子空间语音增强方法推广到多个麦克风的情况,并结合了波束形成算法.实验结果表明,基于信号子空间的麦克风阵语音增强算法消噪后语音的波形失真很小,信噪比提高多. 相似文献
9.
语音信息在竞争情报领域的应用研究已逐步引起学界的注意,在对语音信息进行分类的基础上,从竞争情报过程的角度分析如何构建一个基于企业语音信息的竞争情报提取模型以及各阶段相关的技术,并着重探讨了语音信息挖掘和顸处理技术如何应用于语音竞争情报的提取.该研究试图从一个系统的角度讨论语音信息在收集、存储、处理和语音竞争情报提取方面的总体模式,为今后进一步使用语音竞争情报的探索奠定基础. 相似文献
10.
语音增强可以抑制背景噪声,提高背景噪声条件下语音质量,利于后续语音信号处理。在简要介绍美国军用标准——增强混合激励线形预测语音编码中语音增强算法的基础上,基于TITMS320VC5509a芯片设计一款语音增强模块。详细论述了语音增强算法实时实现的硬件设计、软件优化和开发的关键技术。实际场合实验表明,该模块具有较好的抗背景噪声功能,即可以以独立前置模块的方式用于语音处理系统,也可以以嵌入模块的方式用于语音处理系统。 相似文献
11.
Tomas A. Lipinski Elizabeth A. Buchanan Johannes J. Britz 《Ethics and Information Technology》2002,4(2):143-158
This article explores recent developments inthe regulation of Internet speech, inparticular, injurious or defamatory speech andthe impact the attempts at regulation arehaving on the `body' in the sense of theindividual person who speaks through the mediumof the Internet and upon those harmed by thatspeech. The article proceeds in threesections. First, a brief history of the legalattempts to regulate defamatory Internet speechin the United States is presented; a shortcomparative discussion of defamation law in theUK and Australia is included. As discussedbelow, this regulation has altered thetraditional legal paradigm of responsibilityand, as a result, creates potential problems forthe future of unrestricted and even anonymousspeech on the Internet. Second, an ethicalassessment is made of the defamatory speechenvironment in order to determine which actorshave moral responsibility for the harm causedby defamatory speech. This moral assessment iscompared to the developing and anticipatedlegal paradigm to identify possible conformityof moral and legal tenants or to recognize theconflict between morality and law in assigningresponsibility to defamatory actors. Thisassessment then concludes with possiblesuggestions for changes in the legal climategoverning the regulation of defamatory speechon the Internet, as well as prediction of theresult should the legal climate continue todevelop on its present course. This is not tosuggest that all law, or even the law ofdefamation, be structured to reflect thesubjectivity of a moral construct, but since itis the authors position that the legalassignment of liability in online settings ismisaligned, this reflection can serve asbeginning reassessment of that assignment. 相似文献
12.
本文首先讨论语音信息收集的两种独特来源,接着对建立企业语音数据库的可行性进行分析并提出一种可能的构建模型,在此基础上,探讨如何利用先进的语音处理技术从企业语音数据库中提取竞争情报.该研究试图从一个系统的角度讨论语音信息在收集、存储、处理和语音竞争情报提取方面的总体模式,弥补目前的研究空白,并为今后进一步的研究奠定基础. 相似文献
13.
What is so bad about Internet content regulation? 总被引：1，自引：0，他引：1
John Weckert 《Ethics and Information Technology》2000,2(2):105-111
Legislation was recently introduced into theAustralian parliament to regulate the Internet. Thiscreated a storm of protest from within the computerindustry, where arguments against the legislationranged from those based on technical difficulties tothose based on moral considerations, particularly offreedom of speech and freedom to access information.This paper is primarily concerned with the moralaspects of Internet regulation, but within theparameters of current technology. It will argue thatsuch regulation can be justified, despite the factthat given the current technology there will bedifficulties with enforcement, and reduction inInternet performance. 相似文献
14.
文章提出了一种基于功率倒谱和结合两种高阶统计量——偏度和峰度的语音回声隐藏盲检测方法.根据回声隐藏时对载体语音引入回声的特点,通过分析语音的功率倒谱,提取并结合语音功率倒谱的高阶统计量——偏度和峰度.通过对原始语音样本和藏密语音样本功率倒谱偏度-峰度值的学习和训练,提取得到偏度-峰度阈值,可用于回声隐藏的盲检测分析.实验结果表明,该方法能够对四种回声隐藏方法同时生效,且检测率较高. 相似文献
15.
16.
信息隐藏模型与数字水印技术 总被引：2，自引：0，他引：2
信息隐藏，作为信息安全的新兴领域，是在信息技术特别是网络技术和多媒体技术等迅速发展和广泛普及的基础上发展起来的。本文较系统地论述了信息隐藏模型和数字水印技术，探讨了数字水印技术的嵌入、恢复与检测模型，最后指出数字水印技术的发展前景。 相似文献
17.
Jessica Litman 《Ethics and Information Technology》1999,1(3):213-225
For commercial purveyors of digital speech, information and entertainment, the biggest threat posed by the Internet isn't the threat of piracy, but the threat posed by free speech -- speech that doesn't cost any money. Free speech has the potential to squeeze out expensive speech. A glut of high quality free stuff has the potential to run companies in the business of selling speech out of business. We haven't had to worry about this before, because speaking in a meaningful way to a large audience was expensive, and people couldn't afford to do serious mass speaking for free for very long. The Internet has made it much cheaper. It doesn't take much to give out information to the whole world, every day, for free, for years. And people do. If we are trying to increase the abundant dissemination of information, free speech is good. If we are trying to increase commerce in information, free speech is arguably bad, in that it competes with pay speech. Information merchants would obviously prefer that the only speech in the marketplace be pay speech. In the past two years, commercial content owners have scored significant progress in herding free speakers off the Net. There's an important synergy between persuading the government to give your industry some friendly new laws or regulations, and using new and old legal tools to make life more difficult or expensive for inconvenient competitors who aren't necessarily doing anything illegal. Recently, businesses have been able to combine the two strategies to make the Internet a much safer place to sell speech, by making the Internet a less friendly, more dangerous place to give away speech for free. 相似文献
18.
提出了一种基于感知域的鲁棒性语音认证算法,将语音的感知特性与签名算法相结合,在满足内容认证和身份认证的同时,能够有效地抵抗通信噪声微扰.算法基于语音的掩蔽效应和非线性效应等人耳感知特性,着重去除其时频域掩蔽阈值下的冗余信息,进行非线性滤波后提取感知参数,并运用改进的Rainbow算法对这些语音参数进行签名.实验证明,该算法的唯一性和针对通信噪声的鲁棒性都很好,兼有Rainbow签名的安全性保证,可以满足语音通信中的鲁棒性认证要求. 相似文献
19.
对结构化文档的字体视觉冗余进行研究,利用CSS修改字体外观实现信息嵌入。该思路将加密和隐藏结合,载体应用范围广,隐藏密度大,安全性高,易于软件实现,所提出的嵌入方式对格式文本信息隐藏都具有指导意义。 相似文献
20.
目前的信息传送绝大部分是基于互联网开展的,而由TCP/IP理论构建的互联网从机理上就无法保证传送的信息安全。为此,本文尝试采用数值逼近的经典理论-拉格朗日插值方法,展开数值信息隐藏与分析地研究。通过对传送数据加入前向干扰项、后向干扰项以及隐藏区间的方法,实现真实数值信息隐藏在海量的传送数据之中。不采用本文专有的信息分析方法,通过海量传送数据是无法还原出真实数值信息的。 相似文献