基于分数阶滤波器的ADPCM预测误差信号处理  

ADPCM Prediction Error Processing Based on a Fractional-order Filter

作　　者：宋毅珺[1] 朱艳萍[1,2] 宋耀良[2] 

机构地区：[1]南京信息工程大学电子与信息工程学院,江苏南京210044 [2]南京理工大学电子工程与光电技术学院,江苏南京210094

出　　处：《电声技术》2010年第5期52-55,66,共5页Audio Engineering

摘　　要：ADPCM音频压缩算法在G.721,G.723和G.726等相关数字语音通信协议中得到广泛应用。针对ADPCM中普遍存在的预测误差信号镜频干扰和高频噪声的问题,提出了基于分数阶积分器的误差信号频谱压缩滤波迭代算法和基于分数阶积分器的逆系统信号恢复迭代算法,为进一步提高音频压缩编码品质提供了新的方法,理论分析和仿真结果表明所提出方法的可行性。The ADPCM algorithm of speech compression have been widely applied in digital speech communication protocols, such as G.721, G.723, G.726, etc. An iterative algorithm based on spectrum compression of differential signal with fractional-order integrator and signal reconstruction with its inverse system is proposed, in order to solve the universal problems lied in both the spectrum extension of predicted differential signal and in the noise and mirrorfrequency interferences within high frequency. A new and feasible method is proposed to further improve the audio quality, whose feasibility can be demonstrated by both theoretical analysis and emulation results.

关 键 词：语音信号处理 ADPCM 分数阶滤波 逆系统 

分 类 号：TN912[电子电信—通信与信息系统]