摘要
随着研究的深入 ,语音信号的非线性特征逐渐被人们发现。传统的基于分段线性的语音信号处理方法存在局限性 ,因此用非线性方法对语音信号进行处理对于提高处理质量相当重要。文章介绍了非线性理论在语音信号处理中的一些应用。首先采用延时相图法重构语音的混沌吸引子 ,然后用分形维数、Lyapunov指数等非线性动力学参数反映语音信号的非线性特征 ,设计计盒维数算法并运用于语音分割 ,最后根据语音信号时域波形的分形结构设计了分形内插语音编解码算法。
With the development of research,the nonlinearity of speech is gradually noticed.Conventional speech processing methods based on the short term linear processing limit their power for greater achievement.Therefore,it is important to process speech signals with nonlinear processing methods.In this paper,research on the application of nonlinear processing methods in speech coding and speech recognition is given.First, a chaotic attractor in phase space with the technique of two dimension time delays is reconstructed,and the nonlinearity of speech signals using the characteristics of nonlinear dynamics,i.e.,fractal dimension,Lyapunov exponent is discussed.Then,an algorithm of box counting fractal dimension and its application to speech segmentation is designed.Finally,as speech waveform is fractal,a fractal interpolation algorithm to speech reconstruction is therefore put forward.
出处
《解放军理工大学学报（自然科学版）》
EI
2000年第2期11-17,共7页
Journal of PLA University of Science and Technology(Natural Science Edition)
