论文部分内容阅读
由于在多脉冲激励线性预测声码器中要一次得到所有脉冲的位置和幅度是非常困难的,因此次优的序贯搜索方法被采用,即一次得到一个脉冲。用Matlab仿真观察合成语音效果是可以接受的。
Since it is very difficult to get the position and amplitude of all pulses at once in a multi-pulse excited linear predictive vocoder, a suboptimal sequential search method is used, one pulse at a time. Using Matlab to simulate synthetic speech effects is acceptable.