文档下载 文档下载导航
设为首页 | 加入收藏
搜索 请输入内容:  
 导航当前位置: 文档下载 > 所有分类 > An Introduction tO Hidden Markov Models
侵权投诉

An Introduction tO Hidden Markov Models

The basic theory of Markov chains has been known to mathematicians and engineers for close to 80 years, but it is only in the past decade that it has been applied explicitly to

An Introduction tO Hidden Markov Models L. R. Rabiner

B..H. Juang The basic theory of Markov chains has been known to mathematicians and engineers for close to 80 years, but it is only in the past decadethat it has beenapplied explicitly to problemsinspeech processing. One of the major reasons why speech models,based on Markov chains,have not been developed until recently was the lack of a method for optimizing the parameters ofthe Markov modelto matchobservedsignal patterns.Such a method was proposed in the late1960'sand was immediately applied to speech processing in several research institutions. Continued refinementsin the theoryand implementationofMarkov modelling techniqueshave greatly enhanced the method, leadingtoawide range of applications of these models. It is the purpose of this tutorial paper to give an introduction to the theoryof Markov models,and to illustrate how they have been applied to problems in speech recognition.

Consider next a somewhat more complicated signal— namely a sinewave imbeddedin noise. The noise components of the signal make the modelling problem more complicated because in order to properly estimate the sinewave parameters (amplitude, frequency, phase) one has to take into account the characteristics of the noise component. In the above examples, we have assumed thesinewave partof thesignal was stationary—i.e.nottimevarying. This may not be a realistic assumption. If, for example, the unknown process produces a sinewave with varying amplitude, then clearly a non-linear model, e.g. amplitudemodulation, may be more appropriate. Similarly, if we assume that the frequency, instead of the aniplitude, of the sinewave is changing, a frequency-modulation model might be most appropriate. Linearsystemmodels

INTRODUCTION

YOU ARE GIVEN the following problem. A real world process produces asequence ofobservable symbols. The symbols could bediscrete (outcomes ofcoin tossing experiments, characters from a finite alphabet, quantized vectors from a codebook,etc.) or continuous (speechsamples, autocorrelation vectors, vectors of linear predictioncoefficients,etc.). Your job is to build a signal model that explains and characterizes the occurrence of the observed.symbols. If such a.signal model is obtainable, it then can be used later to identify or recognize other sequences of observations. In attacking such a problem, some fundamental decisions, guided by signal and system theory, mustbe made. For example, one must decide on the form of the model, linear or non-linear, time-varying or time-invariant, deterministic or stochastic. Dependingon these decisions, as well as other signal processing considerations, several possible signal models can be constructed. To fix ideas, consider modellinga pure sinewave. If we have reason to believethat the observed symbols are from a puresinewave, then all that would need to be measured and perhaps phase ofthesineis theamplitude,frequency wave and an exact model, which ex

plains the observed symbols, would result.

A

SSUME

The concepts behind the above examples have been well studied in classical communicationtheory. The variety and types of real world processes, however, does not stop here. Linear system models, which model the observed symbols asthe output of a linearsystem excited by an appropriate source, form another important class of processes for signal modeling and have proven usefui for a wide variety of applications For example,"short time" segments of speech signals can be effectively modeled as the output of an all-pole filter excited by appropriate sources with essentially a flat spectral envelope. Thesignal modeling technique, in this case, thus involves determination of the linear filter coefficients and, in some cases, theexcitationparameters.Obviously, spectral analyses of other kinds also fall within this category. One can further incorporatetemporalvariations of the signal into the linear system model by allowingthe filter coefficients,or theexcitationparameters, to change with time. In fact, many real world processes cannot be meaningfully modeled without considering such temporal variation. Speech signals are one example of such processes. There. are several ways to address the problem of modeling temporalvariation of a signal. As mentioned above, within a"short time" period, some physical signals, such as speech, can be effectively modeled by a simple lineartime-invariant systemwith the 0740-7467/86/O100-0004$O1.OO©1986 IEEE

4

IEEE ASSP MAGAZINE

JANUARY 1986

An Introduction tO Hidden Markov Models

第1页

你可能喜欢

热门文档

相关文档

站点地图 | 文档上传 | 侵权投诉 | 手机版
本站所有资源均来自互联网,本站只负责收集和整理,均不承担任何法律责任,如有侵权等其它行为请联系我们.
文档下载 Copyright 2013 doc.guandang.net All Rights Reserved.  -  email