重庆方言语音识别系统的设计与实现

语音识别赋予了计算机能够识别出语音内容的功能,是人机交互技术领域的重要研究内容。随着计算机技术的发展,语音识别已经得到了成熟的发展。但是关于方言的语音识别还有很大的发展空间。中国是一个幅员辽阔、人口众多的国家,因此方言种类繁多,其中有3000多万人交流使用的重庆方言就是其中之一。采集了重庆方言的部分词语的文本文件和对应的语音文件建立语料库,根据重庆方言的发音特点,选取重庆方言的声韵母作为声学建模基元,选取隐马尔可夫模型(Hidden Markov Model, HMM)为声学模型设计了一个基于HMM的重庆方言语音识别系统。在训练过程利用语料库中训练集语料对声学模型进行训练,形成HMM模型库;在识别过程利用语料库中的测试集语料进行识别测试。实验结果表明,该系统能够实现重庆方言的语音识别,并且识别的正确率为100%。     

Projective synchronization of hyperchaotic system via periodically intermittent control

We further study the projective synchronization of a new hyperchaotic system. Different from the most existing methods, intermittent control is applied to chaotic synchronization in the present paper. We formulate the intermittent control system that governs the dynamics of the projective synchronization error, then derive the sufficient conditions for the exponential stability of intermittent control system by using Lyapunov stability theory, and finally establish the periodically intermittent controller according to the stability criterion by which the projective synchronization is expected to be achieved. The analytical results are also demonstrated by several numerical simulations.     

Projective synchronization of a hyperchaotic system via periodically intermittent control

We further study the projective synchronization of a new hyperchaotic system. Different from the most existing methods, intermittent control is applied to chaotic synchronization in the present paper. We formulate the intermittent control system that governs the dynamics of the projective synchronization error, then derive the sufficient conditions for the exponential stability of intermittent control system by using the Lyapunov stability theory, and finally establish the periodically intermittent controller according to the stability criterion by which the projective synchronization is expected to be achieved. The analytical results are also demonstrated by several numerical simulations.     

基于SPIHT的图像加密与压缩关联算法

为了研究图像压缩与加密同步进行问题, 本文提出了一种在变换域下的图像加密与压缩关联算法在该算法中, 加密过程发生在小波变换与SPIHT编码之间它充分利用了离散小波变换和基于层次树的集合划分(set partitioning in hierarchical trees, SPIHT)编码属性, 扩散过程被限制在单个子带内部. 此外, 混淆过程保留了SPIHT编码中两个最重要的位和符号位, 它包含了图像的重要信息. 实验结果表明, 算法具有良好的安全性、图像重构视觉质量以及很高的加/解密速度.