一种有限误差线谱频率参数暂时分解量化算法

本文对线谱频率参数量化提出了一种有限误差的暂时分解算法.新算法以原有的暂时分解算法为基础,重新定义了误差约束条件下的插入事件产生规则,并引入了事件消除机制.实验表明,在相同的编码速率下与原算法相比,该算法在对线谱频率参数量化过程中能够取得更低的平均谱失真度,从而获得更高的语音质量,更适用于低速率语音编码系统.

Limited error temporal decomposition algorithm for line spectral frequency parameters quantization

In this paper,a limited error temporal decomposition algorithm is presented for the line spectral frequency parameters quantization.Based on the original temporal decomposition algorithm.the new algorithm redefines the generation rule of the inserted event with the error constraint and introduces an event elimination method.The results show that compared with the original algorithm,the new algorithm can achieve a lower average spectral distortion for the line frequency spectrum parameters quantization in the same encoding rate condition to get the higher speech quality,and is more suitable to apply in the low bit-rate speech coding system.