为在极低速率下实现高质量的语音编码,提出了一

种新的有效的线谱对( LSP)参数量化算法——P-RS- MSM Q

算法。此算法以多帧联合矩阵量化作为基本框架,引入了基

于超级帧模式的均值去除和帧间预测策略、 矩阵分裂和子矩

阵多级量化策略 同时提出了基于语音帧短时谱能量的帧

内加权和基于超级帧中各子帧重要性的帧间加权策略等。-Very low rates in high-quality speech coding, a new and effective LSP (LSP) parameter quantization algorithm- P-RS-MSM Q algorithm. This algorithm to quantify the multi-frame as the basic framework of the joint matrix, the introduction of super-frame model based on the mean removal and interframe prediction strategy, matrix splitting and multi-level quantization matrices　the same time frame was proposed based on short-term spectrum of speech frame energy weighted and based on the sub-frame super-frame, the importance of inter-weighted strategies.