周瑾, 乔树山, 凌康, 詹毅, 蒋见花. 基于G.729的快速固定码书搜索算法[J]. 微电子学与计算机, 2020, 37(11): 13-16,23.
引用本文: 周瑾, 乔树山, 凌康, 詹毅, 蒋见花. 基于G.729的快速固定码书搜索算法[J]. 微电子学与计算机, 2020, 37(11): 13-16,23.
ZHOU Jin, QIAO Shu-shan, LING Kang, ZHAN Yi, JIAN Jiang-hua. A new method of fast fixed-codebook search algorithm on G.729 coding[J]. Microelectronics & Computer, 2020, 37(11): 13-16,23.
Citation: ZHOU Jin, QIAO Shu-shan, LING Kang, ZHAN Yi, JIAN Jiang-hua. A new method of fast fixed-codebook search algorithm on G.729 coding[J]. Microelectronics & Computer, 2020, 37(11): 13-16,23.

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

  • 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.

     

    Abstract: In Conjugate-Structure Algebraic-Code Excited Linear-Prediction (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.

     

/

返回文章
返回