| 基于GERBF神经网络非线性预测的ADPCM算法控制系统中的应用 |
| |
| 引用本文： | 李志宏,齐向东,王安红.基于GERBF神经网络非线性预测的ADPCM算法控制系统中的应用[J].太原科技大学学报,2004,25(3):181-184. |
| |
| 作者姓名： | 李志宏 齐向东 王安红 |
| |
| 作者单位： | 太原重型机械学院电子信息分院,太原,030024 |
| |
| 摘 要： | 用神经网络建立非线性预测模型对语音信号进行处理,采用后向预测建模,不增加传输码率。采用一种改进的广义径向基函数网络(GERBF),利用正交最小二乘法训练速度快的优点,降低算法的复杂性。实验表明:基于GERBF预测器的语音编码系统在嵌入维数很少时亦能较好地去除语音信号相关性,其恢复语音质量优于CCITT,建议G.721中的ADPCM算法。
|
| 关 键 词： | 语音编码 非线性预测 GERBF网络 正交最小二乘法 |
| 文章编号： | 1000-159X(2004)03-0181-04 |
| 修稿时间： | 2004年4月20日 |
A Nonlinear Prediction ADPCM Algorithm Based on GERBF Neural Network |
| |
| LI Zhi-hong,QI Xiang-dong,WANG An-hong.A Nonlinear Prediction ADPCM Algorithm Based on GERBF Neural Network[J].Journal of Taiyuan University of Science and Technology,2004,25(3):181-184. |
| |
| Authors: | LI Zhi-hong QI Xiang-dong WANG An-hong |
| |
| Abstract: | A nonlinear prediction speech coding ADPCM algorithm based on RBF is introduced in this paper. In this coding system, the bit rate is not increased by the backward predictor. An improved General Radical Basis Function neural network (GERBF), which is trained by rapid Origin Least Square (OLS), is proposed. Compared with ADPCM algorithm based on BP, the computation is less evidently. Besides, the speech quality of the new system has an improvement of 1-2 dB than that of the speech coding standard G.721 system. |
| |
| Keywords: | speech coding nonlinear prediction GERBF network OLS |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
