SIGNAL PROCESSING

Generally, all speech signals produced contain some form of noise constituent. The noise may be due to the inaccuracies involved in writing code of the transmitted waveform, due to the addition of acoustically coupled background noise or the source of the noise can be from the instruments used such as thermal and electrical noise. In apparent randomness of these signals, they are usually carrying useful information concerning the physical variations. The major task in digital signal processing is to extract the useful and desired information from the corrupted signals. Estimation techniques is required for smoothing, filtering, or prediction and estimation of models for signal generation such as autoregressive models. Under such circumstances, there is a need to work on the design of more robust filters, such as adaptive filters which tracks the changes of signal and noise.