查询词典 hmm

During that period, his research field covered many sections, such as speech analysis and synthesis, speech coding, automatic answering, speech recognition, echo elimination, sound operation interface, high efficiency search algorithm and identifiable HMM.

在贝尔实验室期间，他的研究领域覆盖了语音分析与合成，高音质、中低比特率语音编码，自动应答，语音识别，回声消除，无手控语音操作界面，高效搜索算法，有辨识力的HMM训练等等多个方面。

The AR spectrum is not restricted by length of data, and AR spectrum parameters are sensitive for law of condition change. Autoregressive transformation is made to vibration signals, and then AR spectrum coefficients are got which can be utilized as feature vectors. Fault diagnosis method for centrifugal pump Based on AR and 2D-HMM is produced in this paper. The topology of this model and its parameters were introduced too.

利用AR谱不受数据长度的限制，AR模型参数对状态变化规律反映敏感的特点，以振动信号做自回归变换后的AR谱系数作为特征向量，将基于AR的2D-HMM引入到离心泵故障诊断中，提出了一种基于AR的2D-HMM故障诊断方法，并论述了该模型的拓扑结构和主要参数以及相应的训练和识别算法。

VQ is easy and tend to realize, HMM can comminute the speech dynamically, it can embody not only the dynamic characteristic but also the stationary characteristic of the speech. GMM is a special HMM, it uses combination of many weighted Gaussian density function to approach the distributing density of speaker characteristic vector in characteristic space, so it is also a good way.

VQ方法是最简单又易于实现，HMM方法能将语音动态分割，既能体现语音的动态特征，又能体现语音的平稳特性，GMM用多个加权高斯密度函数的组合来逼近说话人特征矢量在特征空间的分布密度，收到了较好的结果。

One of the key techniques in Continuous Handwritten Character Recognition is how to model every word entry in lexicon and every sentence that made up by the words. Due to the HMM′s characteristic of time sequence modeling capability, a Cascaded Hidden Markov Models, which defines model connection probability and state transition probability between HMMs, is proposed in this study.

利用HMM对时间序列的较强的建模能力这一特点，提出了可用于连续字符识别的HMM级联模型；给出了字符HMM模型连接概率和模型间状态转移概率的定义，并通过修正训练算法的重估公式，重估字符模型的连接参数，描述了用于手写体识别的字符HMM的设计方案；给出了级联训练算法重估公式和级联Baum-Welch训练算法描述。

Up to this time, the major challenge to sign language recognition is how to develop approaches that scale well with increasing vocabulary size In this paper, an approach to large vocabulary, continuous Chinese sign language recognition is presented, which uses subwords instead of whole signs as the basic units Since the number of subwords is limited, HMM based training and recognition of the CSL signal become more tractable and have the potential to recognize enlarged vocabularies Furthermore, the proposed method facilitates the CSL recognition when finger alphabet is blended with gestures About 2400 subwords are defined for CSL One HMM is built for each subword, and then the signs are encoded based on these subwords A decoder that uses tree structured network is presented Clustering of the Gaussians on the state, language model and N best is used to improve the performance of the system Experiments on a 5119 sign vocabulary are carried out, and the correct rate is over 90% for continuous sign recognition

迄今为止，手语识别面临的最大问题是如何解决词汇集易扩充的连续识别提出一种大词汇量连续中国手语识别方法，将词根作为识别基元，由于基元的数目是有限的，因此基于HMM的手语信号的训练和识别变得比较容易处理，可以实现更大词汇量的识别除此之外，所提方法还有利于实现手势语和手指语的混合识别从中国手语中共整理出2 4 0 0多个词根，为每个词根建一个并行的HMM模型，对各数据流的HMM模型进行聚类，确定出手语识别的基元根据这些基元对手势词编码，并建立了树状搜索网格，使用状态结点上高斯密度函数聚类、语言模型和N Best方法提高系统的速度和精度对 5 119个手语词做了实验，连续语句的识别率可在 90 %以上

HMM is good at dealing with sequential inputs,while SVM shows superior performance in classification especially for limited samples.Therefore,they can be combined to get a better and effective multilayer architecture classifier.SVM is used to resolve the uncertainty of the remaining signal which is confusable after the HMM-based recognition.

针对战场环境下声信号的特点，算法综合考虑HMM适合处理连续动态信号及SVM小样本情况下的强分类能力，利用HMM处理待辨识的连续动态信号，将HMM易混淆的信号作为与待辨识信号较为相似的模式类，形成候选模式集，再由SVM在候选模式中对待辨识信号作最后决策。

Then, studying the theories and algorithms of Hidden Markov Models in detail, and discusses some issues in actual applications and gives corresponsive solution means to them. At last the functions of HMM used in cutting vibrations forecast are given.

第三章应用HMM中的Hterbi算法进行信号的数值滤波处理，讨论了振动信号的特征提取方法，以及幅值谱矢量的标量量化技术，最后提出了机床切削过程中的FFT-HMM颤振预报方法。

This paper bases on the former documents which related to the distance in HMM (we also proved Juang and Rabine's intuitionistic discussion strictly ), and study the distance in HMM from two aspects:the one is to find the more rapid and effective approximate algorithm, the other is to introduce a new and more applied distance.

针对J-R距离在理论上很有意义，但实际上难以计算且结果具有不确定性的缺点，利用文献[11]，得到了近似计算离散HMM和具有混合高斯观测密度的连续HMM距离的一种快速有效的算法，该算法用矩阵形式来表达，便于用MATLAB进行计算。

By using a tree pictorial structure model to represent the detuned generic people in the video, and a modified Hidden Markov Model to simulate the motion of people between the two frames of the video, a machine learning method is used to the modified HMM to obtain the estimation of parameter of the modified HMM,and to capture the people model from the video.

利用片图模型表示未经学习的人体，改进的隐马尔可夫模型模拟人体在视频序列各帧间的运动，并使用机器学习方法对该改进的HMM进行推理，获取改进HMM的参数，从而获得所需的人体模型。

In the training of HMM, referring to the auxiliary function in reference〓, and the method in reference〓 to get the estimations of the parameters of Markov chain, we define the likelihood function and auxiliary function for the maximum likelihood estimation of CDHMM, and then induce systematically the forward-backward algorithm and the re-estimation formulas for parameters of CDHMM.

在HMM的建模方面，本文借鉴文献〓中求多变量观测值马尔可夫链参数时辅助函数的选取方法，定义了CDHMM的似然函数和辅助函数；借鉴文献〓中求解马尔可夫链参数时的求偏导方法，并针对本文辅助函数加以简化和改进，系统地推导了HMM的前向—后向算法，以及CDHMM各参数的极大似然估计公式。

Kate: I'm sorry I kissed you.

抱歉，我亲了你。

This means giant network to buy long swim the hearsay of the net is complete by checkmate.

这意味着巨人网络收购久游网的传闻彻底被击破。