音频信号联合编码的通用码书研究

Research on General Codebook of Joint Coding Algorithm for Audio Signal

摘要:

为了减小音频信号编码算法的运算量,提出了在线性预测编码、SOM神经网络矢量编码以及 Huffman编码相结合的声音信号联合编码算法(以下简称联合编码)的基础上,将算法中原有的专用码书改为通用码书的算法。利用 Matlab软件编程进行了专用码书和通用码书条件下的声音信号编解码实验。实验结果表明,在码率相等的条件下,使用通用码书和专用码书的联合编码方法可以得到相近的译码声音质量。使用通用码书的联合编码的最低码率在音频编码格式 Opus码率的范围内,且接近 Opus码率的下限,而且算法较 Opus编码简单,因而实时性较好。文章提出的编码算法可为音频压缩编码的进一步研究提供参考。

Abstract:

In order to reduce the computational complexity of the audio signal coding algorithm, in this paper, the algo.rithm was put forward which was based on the combination of linear predictive coding, SOM neural network vector codingand Huffman coding, and the original special codebook was changed to general codebook. Using Matlab software program.ming, the codec experiment under the condition of special codebook and general codebook was carried out. The experimen.tal results show that under the condition of equal code rate, the joint coding method of universal codebook and special co.debook can get similar decoding sound quality. The minimum code rate of the joint coding using the general code book iswithin the range of the code rate of the audio coding format Opus and is close to the lower limit of the Opus code rate, andthe algorithm is simpler than the Opus coding, thus the real-time performance is better. The coding algorithm proposed inthis paper can provide reference for further research of audio compression coding.