摘要
将小波变换和非线性动力学研究方法相结合,分析研究了在微振动语音信号检测中两类干扰背景的特性及其对检测结果的影响,讨论了两类情况下的解决方法。先通过小波分解得到原始时间序列的低频和高频部分,再分别对其进行处理。结果表明,对随机噪音干扰信号而言,可以通过放大低频、抑制高频来增强微弱信号;而对于强混沌干扰背景信号,混杂信号低频部分和干扰背景的奇异吸引子最相似,因此需要通过抑制低频、放大高频来增强微弱信号。最后通过Matlab音频仿真证明了该结论。
By combining the methods of wavelet transform and nonlinear dynamics, the properties of two sorts of interference background and their influences upon the results in weak vibration speech signal detection are analysed. The original data series is decomposed into one low-frequency part and several high-frequency parts by wavelet transform and they are processed respectively. The results show that , to enhance the weak signal, we need to magnify the low-frequency part and suppress the high-frequency parts if the interference background is random noise, while for chaotic background, we need to suppress the low-frequency part and magnify the high-frequency parts because the attractor of the low-frequency part is most similar to the chaotic background's. Finally we certify this conclusion using Matlab audio frequency simulation.
出处
《计算机与数字工程》
2009年第3期82-84,90,共4页
Computer & Digital Engineering
关键词
小波变换
混沌背景
相空间重构
奇异吸引子
wavelet transform, chaotic background, phase space reconstruction, strange atrractor