Fundamentals of Linear Prediction

The linear prediction method provides a robust, reliable and accurate method for estimating the parameters that characterize the linear time-varying system. It is very important tool in digital signal processing[11] because it deals with applications in variety of areas such as speech signal processing, image processing etc. It has become the predominant technique for estimating the basic speech parameters such as pitch, formants, and spectra. The importance of linear prediction lies in it’s ability to provide extremely accurate estimates of speech parameters as well as it’s relative speed of computation. The proposed talk covers the linear prediction model which deals with the problem of predicting the values of a stationary random process either forward in time or, backward in time known as forward and backward prediction. I will discuss basic principles of linear prediction analysis and the fundamental problem related with the analysis as applied to speech processing. As linear prediction refers to a variety of formulations of the problem of modeling the speech waveforms, I will discuss few basic formulations of linear prediction analysis and examine the similarities and difference among them. I will then highlight the basic problems related with linear prediction as applied to speech which will lead to PLP analysis, an extension over linear prediction.