LSF标量格型混合量化  

Scalar and Lattice Hybrid Quantization of LSF

作　　者：潘旭遥[1] 宋国新[1] 

机构地区：[1]华东理工大学计算机科学与工程系,上海200237

出　　处：《华东理工大学学报（自然科学版）》2006年第9期1086-1090,共5页Journal of East China University of Science and Technology

基　　金：上海市科技发展基金(045115009)

摘　　要：为了降低线谱频率(LSF)参数矢量量化器的搜索复杂度和码字存储单元,利用格型矢量量化的优点,设计了一种适合LSF参数量化的标量格型混合量化器。该量化器对LSF参数的预测残差矢量的第一、二个参数进行标量量化,余下的参数则利用格型矢量量化,从而降低了搜索复杂度和码字存储单元,与G.729协议所使用的LSF量化技术相比,有一定的改进。To reduce the computational complexity and storage of the line spectrum frequency（LSF） parameters quantizer, a scalar and lattice hybrid quantizer is designed to quantize LSF parameters based on the attractive properties of lattice vector quantization（LVQ）. The first and second components of the LSF predictive error vector are quantized by scalar quantization, but the rest components are quantized by LVQ, which takes advantage of lattice＇s properties to effectively reduce the complexity of the design and search of the codebook and also decrease the storage. Experimental results show that this quantizer has comparable performance with the LSF quantization method used by G. 729.

关 键 词：语音编码 线谱频率 标量量化 格型矢量量化 

分 类 号：TN912.3[电子电信—通信与信息系统]