Input signal through a first-order high-pass
optical filter filtered DC component, and then be divided into four 60-value of the sub-frame, each frame of an independent analysis of LPC. In order to improve the continuity of LPC coefficient, using a length of 180 samples of the overlapping windows that contain both before and after the two sub-frame, which makes algorithm to introduce value-60 ahead of the delay, the algorithm of the total delay of 37. 5 milliseconds. LPC coefficient of linear spectrum frequency (LSF) said, LSF parameters used to split VQ forecast, only the fourth sub-frame. In order to improve the quality of quantitative sense, high pass filter after the voice signal to be perceived by Formant weighted filter harmonious Zhenfeng noise shaping filter to generate initial target signal. The former sub-frame by the parameters of the LPC did not quantify a factor, the latter two on each sub-frame for open-loop estimated by Pitch, Pitch the range of 18 to 142 samples value. LPC synthesis of
optical filter, filter resonance-weighted perception of a harmonious Zhenfeng noise shaping filter system is used to enter the best response to the calculation and incentive estimates. G. 723.1 encoder also includes a five-order pitch forecast, according to its open-loop pitch parameter estimates and the impulse response to the closed-loop pitch searchable. In the best incentive is estimated, from the initial target signal to be in the minus zero input system response and contribution to the pitch predictor signal to be the ultimate goal, and then were used for high and low bit-rate MP-MLQ and ACELP methods to quantify. One LSF parameters, pitch value and incentive parameters to be transmitted to the decoder. First of all decoders in accordance with the parameters of the LSF reconstruction of LPC filter, then pitch value and encourage the parameters are adaptive code-excited signal and fixed code-excited signal. In order to enhance the reconstruction of the subjective voice quality, decoder also includes a post-filter, filter after the after Formant pitch and filter components. Encouraging signal followed by pitch after the filters, synthetic filters and resonance peak of reconstruction after the
optical filter voice.