| With the development of communication technology and the improvement of the speech transmit technology in real time based on Internet, the high request to transmission efficiency and memory capacity of the speech is needed. One of the main ways to solve these problems is the speech coding.The code-excited linear-prediction(CELP) algorithm is considered one of the most successful algorithm. And lots of famous international standards are based on it. The CELP algorithm can produce high-quality synthetic speech signal at the middle coding rate, but the synthetic speech quality is obviously descent at the coding rate less than 4kbit/s. In order to enhance the quantification efficiency, it is necessary to add the length of the exited vector. The speech quality will debase drastically if giving few pulse to denote the exited signals. This is the reason why it is difficult to enhance the speech quality at the coding rate less than 4kbit/s.This article is devoted to research to optimize the algebraic codebook with lesser bits. On the base of the G729 algorithm, the precession-objection is 20ms-time voice-flame so that the bit-rate is decreased half to 4kb/s. Therefore the algorithm of multiple pulse-dispersion algebraic codebook-excited linear-prediction is developed. Firstly, based on G729 model, a new algebraic structure of fix codebook is devised to reduce the algorithm complexity. And then, multiple pulse-dispersion technique is applied to disperse the vectors of fixed codebook. The multiple pulse-dispersion technique is realized by a low-pass finite impulse response filter which cut-off frequency is 3400Hz.In the end of the article, the algorithm is simulated by the MATLAB program language. The simulated outcome indicates that the output voice wave is similar with the input voice wave. But at the segment of atonic voice there are some distortions and there are some time-delays. Thereby it can be testified that the algorithm is theoretically feasible. |
PDF Full Text Request