语音识别的非线性方法

语音信号是一个复杂的非线性过程，这使得基于线性系统理论发展起来的传统语音识别技术性能难以进一步提高。近年来人们开始逐渐重视非线性在语音识别技术中的应用，本文概括地介绍了非线性理论在语音识别技术中的所取得的成果和发展方向，除了涉及较为流行的隐马尔柯夫过程和人工神经网络在语音识别中的应用外，文中着重论述了近年来发展迅猛的混沌，分形理论在语音识别中的应用，本文最后还提到不可忽视的分形理论在语音编码中的应

Non linear Methods for Speech Recognition

Speech signal is traditionally treated as a linear process.However,it is indicated by extensive research that the speech signals are actually complicated non linear processes.To improve the recognition rate,recent research effort has started to migrate to analyze speech signal using non linear theory.This article summarizes the new development in this area.Besides HMM and ANN,which have been widely used by many authors ,this paper introduces in particular a series of fast growing non linear such as chaotic and fractal theories and their applications in speech recognition and coding .