摘要
虽然G.729中采用的集中搜索和G.729a中采用的深度优先树搜索可以有效减少固定码本搜索复杂度,但固定码本搜索在整个语音编码算法中仍占有较大比重。为了在基本维持语音质量的前提下,减少搜索运算量,研究了几种快速搜索算法,脉冲替代和预选替代一个脉冲搜索算法可以大大减少搜索次数,但语音质量明显下降,因此提出每次替代两个脉冲搜索算法,得到比替代一个脉冲较为完整的搜索,产生较好的语音质量。仿真结果表明,该算法可以大大减少搜索运算量,并且保持了和G.729a深度优先树搜索算法相同的语音质量。
The focused search used in G.729 and depth-first tree search used in G.729a can reduce the fixed-codebook searching complexity efficiently.However it still occupies larger proportion of whole encoding complexity.To reduce the computational complexity of codebook search,several codebook search methods are studied.Pulse replacement search method and pre-select one pulse replacement search method can reduce search load,but speech quality will be worse.In the proposed method,replacing least important pulse considers two pulses instead of one pulse.The simulation shows that the computational complexity of the proposed method is reduced largely and the quality is degraded slightly compared to original depth-first tree method.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第10期2209-2211,2215,共4页
Computer Engineering and Design
基金
河北省教育厅自然科学基金项目(DZ-2005-06-Z)
河北省科技公关基金项目(072135114)
关键词
固定码本
代数码激励线性预测
集中搜索
深度优先树
脉冲替换
fixed-codebook
algebraic-code-excited linear-prediction
focused search
depth-first tree
pulse replacement
