The Optimal Gain Filter of LD-CELP

The recommendation G.728 depends on the Levinson-Durbin algorithm to update gain filter coefficients. In this topic, it is replaced by three different methods which are the weighted L-S recursive filter, the finite memory recursive filter and the BP neural network, respectively. Using these three gain filter the speech coding effect is all better than the G.728. The weighted L-S algorithm has the best result. Its average segment SNR is higher than the G.728 about 0.76dB. It is also used to evaluate the case that excitation vector is 16 and 20 samples respectively; the weighted L-S algorithm has similarly the best result.