码本

At the same time, the practical implementation of evolutionary algorithm is given in order that evolutionary algorithm is introduced into vector quantizer. In the fourth section the asymptotic theory of vector quantization is studied. Then the partial distortion theorem that when codebook size is large enough, each region of the partition makes an equal contribution to the distortion for an optimal quantizer, which is the key part of vector quantizer is obtained. Finally, the competitive learning algorithm based on the partial distortion theorem is proposed, which introduces the subdistortion of each region into the distortion measure to assure that the subdistortion of each region approximately equals to each other.

第四章详细研究矢量量化的渐近理论，然后得出了矢量量化器设计的核心部分——部分失真定理，当码本尺寸足够大且输出矢量的渐近概率密度与输入矢量的概率密度的幂成比例时，最优量化器划分的各个区域对该量化器的平均误差产生相同的影响，最后提出基于部分失真定理的竞争学习算法，它通过在误差测度中，引入与各区域相关的子误差以保证各区域的子误差近似相等，该算法实现较为简单，在一定程度上确保全局最优。

Finally, the design and implementation of vector quantizer are discussed. In the third section the theory of genetic algorithm, evolutionary strategy and evolutionary programming are discussed in detail.

第三章详细讨论了遗传算法，进化策略和进化规划这三种算法的理论，并给出了具体实现，为进化算法引入到矢量量化的码本设计中打下基础。

More detailed content of this dissertation are as follows In the second section the definition of vector quantizer is firstly given. Then the features of conventional codebook design algorithm are briefly presented.

本文的绪论部分对相关技术作了充分的回顾，其后的内容如下：第二章首先介绍矢量量化器的定义，接着简要给出常用的码本设计算法的特点，最后讨论矢量量化器的设计与实现。

The LP analysis is used again on residual signal to reconstruct the spectrum fine structure.Results Adaptive codebook delay and two codebook gains are tied to build pitch/energy contour feature.

结果该方法通过对残差信号进行线性预测分析，提高了谱包络的精确程度，并从码本增益和延时参数中得到了基音/能量轨迹特征。

Further on, the Trifurcate Tree Searching algorithm was proved to have the least complexity, rather than the Binary Tree Searching algorithm, which was conventionally taken as the one with the least complexity in VQ tree searching. The application of ANS to speech enhancement was studied in chapter six.

第五章设计并实现了一种基于矢量量化技术的语音增强系统，采用了一种新的码本分离技术即采用特征矢量中对噪声不敏感部分用于VQ搜索、而用剩余部分用于形成增强后的语音；证明了矢量量化树搜索中三分法复杂性为最小而不是通常认为的二分法。

Though new codebook searching method debases the searching veracity , it scales down data processing quantity of searching and reduces the complexity .

此方法在次最佳码本搜索法的基础上作出了改进，尽管降低了搜索的准确性，但却大大减少了搜索时的数据处理量，降低了算法复杂度。

Chapter 2 covers the AVQ , which has two dynamic codebooks in encoder and decoder, for encoding speech signal.

在第二章，研究了语音信号的自适应矢量量化（或称动态码本的矢量量化）。

After classified by VQ codebooks, the speech signal is put into SVM pairwise classifiers for further processing.

我们将待测语音与VQ码本比较，输出几位最相似的说话人作为候选，然后将待测语音输入到由这几位说话人构建的SVM一对一组合分类模型中，进行细分类。

Based on statistics states for source coding and study of a design of CDMA structured codebooks, the paper also presents the basic set-up of side information-aided at decoder, and approaches a good transmission performances under low-WNR and high-WNR that corresponds to the attack channel ambiguity.

该方法在分析信源编码的统计基础上，通过研究CDMA结构性码本的设计条件，在接收端利用边信息辅助解码，达到了在攻击信道条件下，适应于高低WNR环境的数据隐藏通信模式。

By this step, lengthy codewords with many leading zeros, which are common in the codebook, are represented in a compressed form by the exponent and mantissa fields of the floating point result.

通过此步骤，在码本中常见的，具有多个前导零的长代码字由浮点结果的指数和尾数字段以一种压缩的形式表示。

I didn't watch TV last night, because it .

昨晚我没有看电视，因为电视机坏了。

Since this year, in a lot of villages of Beijing, TV of elevator liquid crystal was removed.

今年以来，在北京的很多小区里，电梯液晶电视被撤了下来。