改进的能量参数预测多级矢量量化算法  

Improved Predictive Multistage Vector Quantization of Energy Parameter

作　　者：肖强[1] 陈亮[2] 

机构地区：[1]解放军理工大学通信工程学院研究生4队,江苏南京210007 [2]解放军理工大学通信工程学院电子信息工程系

出　　处：《军事通信技术》2010年第2期11-14,共4页Journal of Military Communications Technology

摘　　要：能量参数的量化精度对低速率语音编码的语音质量有着重要影响。文中提出了一种基于预测多级矢量量化联合优化码本设计的能量量化算法,采用渐进闭环设计来进行预测器系数和多级矢量量化码本的迭代更新,从而实现预测器系数和量化码本的联合优化。仿真结果表明,与传统的预测多级矢量量化相比,文中算法量化误差降低了28%左右,具有更高的量化精度。The energy quantization precision influences the quality of low bit rate speech coding.An improved predictive multistage vector quantization of energy parameter algorithm was proposed.Joint optimization of both the predictor and the multistage vector quantizer was performed with the asymptotically closed loop design.A jointly optimal design was used to train the codebooks.Compared with the traditional algorithm,the simulation results indicate that the new algorithm can reduce the quantization error by 28% and obtain higher quantization precision.

关 键 词：低速语音编码 预测多级矢量量化 联合优化 

分 类 号：TN91[电子电信—通信与信息系统]