Kalman filtering technique .To obtain a reconstructed speech signal, which is similar to the input signal. To achieve these results, sample speeches were obtained. These wer e modeled as an autoregressive (AR) process and represented in the state space domain by the Kalman filter.
Definition of Filter.
Block diagram:
Classification of Filters.
Analog Filters.
Digital Filters
Its Advantages:
A digital filter is programmable. Easy to design, test and implement. It can even handle low frequency signals accurately.
LMS Filter.
Wiener Filter
Kalman Filter.
Stationary properties of the Kalman filter are not required for the eterministic dynamics or random processes.
DefinitionIt is an estimator for linear quadratic problem which focuses on estimating the instantaneous state of a linear dynamic system perturbed by white noise.
Advantages
It is only a tool. It is a computer program. It is a complete statistical characterization of an estimation problem. In a limited context, it is a learning method.
of estimators.
measurements.
As such, these equations for Kalman filter fall into two groups: Time Update Equations and "Measurement Update Equations
Initial estimator
Responsibilities of time update equations are for projecting forward (in time) the current state i.e to obtain the priori estimates time step Measurement update equations are responsible for the feedback i.e. for incorporating a new measurement into the priori estimate to obtain an improved posteriori estimate Time update equations can also be thought as predictor equations while the measurement update equations can be thought as corrector for the next
equations
The above figure resembles that of a predictor corrector algorithm for solving numerical problems .
Definition.
The measurement noise covariance R is practically possible aid and is measured before the operation of the filter.
The process noise covariance Q should be measured while operating the filter which is a difficult task.
When Q and R are constants, both estimation error covariance Pk and the Kalman gain Kk will stabilize quickly and then remain constant. In this case, these parameters can be pre computed by running the filter off line.
In this thesis, an implementation of employing Kalman filtering to speech processing had been developed. The simulated results had proven that t
The results have also shown that Kalman filter could be tuned to
Moreover, a test for cross correlation had also been conducted during thi
s thesis for measuring the similarity of the input and output speech signal
s. This test is of necessity for the reason that different signals are boun d to be similar but not identical.