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 because it deals with application 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.
In the proposed talk, I will cover the two types of linear prediction
which deal 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 principals 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.
Additional items of interest: