Studying of the code-excited linear prediction of speech coding

SUMMARY Studying of the Code-Excited Linear Prediction of Speech Coding It is possible that transmission of speech signal over the channels which have low capacity by using digital coding techniques and its storage by using a low memory. In this thesis, it was examined a general structure of the speech coding systems with code-excited linear predictive coders. Analysis-synthesis process is generally done for coding of speech signals. While analysis process is implemented at the transmitter, synthesis process is implemented at the receiver. Linear prediction coding technique is used in the analysis-synthesis process. At analysis process, indexes and gains of stochastic and adaptive code tables for the related speech signal that are found with search algorithms. Synthesis of the speech signals sent from same stochastic and adaptive code tables is done by using these indexes and gains at the receiver. In this working subjects as predictor conception, determination of window functions and pitch period, all of which effective on performancing of the speech coding, were investigated with details. Various coding techniques were compared from the point of view amount of process, data rate, signal/noise ration and audial quality. Obtained results show that code-excited linear prediction speech coders have an excellent performance. Key Words: Speech Signals, Speech Coding, Linear Predictive Coefficients, Pitch Period, Linear Prediction. VI