摘要
介绍了隐Markov模型原理,它是用来描述含有未知参数的Markov过程,是描述随机过程统计特性的概率模型。在此基础上,设计了基于HMM模型的孤词检测实验,通过优化实验模型,采用Baum-Welch算法解决HMM模型的训练问题,找到HMM模型估计参数λ值,这在数学角度上等价于其他线性预测系数。此实验在减少不必要的HMM训练的同时,降低了算法复杂程度。为了测试Baum-Welch算法的有效性,进行了数据仿真实验,结果表明该算法是有效的。
This paper introduces the principle of Hidden Markov Model, which is used to describe the Markov process with unknown parameters, is a probability model to describe the statistical properties of the random process. On this basis, designed a solitary word detection experiment based on HMM model, by optimizing the experimental model, Using Baum-Welch algorithm for training the problem of solving the HMM model, HMM model to estimate the parameters of the Z value is found, in this view of mathematics equivalent to other linear prediction coefficient. This experiment in reducing unnecessary HMM training at the same time, reduced the algorithm complexity. In order to test the effectiveness of the Baum-Welch algorithm, The simulation of experimental data, the results show that the algorithm is effective.
出处
《河北科技大学学报》
CAS
2015年第1期52-57,共6页
Journal of Hebei University of Science and Technology
基金
河北省自然科学基金(F2012208018)
河北省高等学校科学技术研究重点项目(ZD2014027)