基于G.729的快速固定码书搜索算法  

A new method of fast fixed-codebook search algorithm on G.729 coding

作　　者：周瑾 乔树山[1,2] 凌康 詹毅[1] 蒋见花 ZHOU Jin;QIAO Shu-shan;LING Kang;ZHAN Yi;JIAN Jiang-hua(Institute of Microelectronics,Chinese Academy of Science,Beijing 10029,China;University of Chinese Academy of Sciences,Beijing 10049,China)

机构地区：[1]中国科学院微电子研究所,北京100029 [2]中国科学院大学,北京100049

出　　处：《微电子学与计算机》2020年第11期13-16,23,共5页Microelectronics & Computer

基　　金：航空科学基金(201743X2002)。

摘　　要：共轭结构代数码激励线性预测编码算法(Conjugate-Structure Algebraic-Code Excited Linear-Prediction, CS-ACELP)的固定码书搜索算法通过嵌套循环计算出码书矢量,这个算法的重要缺陷是嵌套循环深度大,时间复杂度高,计算量大.在重排序搜索(Reordering Search Sequence)算法的基础上,对搜索结构进行改进.将预搜索部分获得的组合脉冲直接用于下一步的全搜索,将嵌套循环的深度由4层降至2层,降低时间复杂度和计算量.并且与固定码书搜索的几个经典算法相比,搜索次数减少了30%到98%不等,同时具有优于G.729A算法的优越编解码音质.In Conj ugate-Structure Algebraic-Code Excited Linear-Predict ion(CS-ACELP)algorithm,the algebraic codebook search calculation is through the nested loops.This leads the algorithm with a large computation,high computational complexity and time complexity.After thoroughly analyzing the reordering search sequence approach*this paper proposes a fast search method for fixed-codebook search based on modification of the search structure.In the proposed method,the combination of pulses obtained in the pre-search are directly used for the next full search Therefore,the depth of nested loops in search is mitigated from 4 to 2.Compared with several other classic algorithms for fixed-codebook search,search times of the proposed method get a decreases of 30% to 98%.This method effectively reduces the computational complexity and time complexity,and codec sound quality is better G.729A.

关 键 词：G.729 固定码书搜索 重排序搜索算法 PESQ 

分 类 号：TP37[自动化与计算机技术—计算机系统结构]