【摘要】 提出一种带反馈单元的动态小波神经网络(DWNN)并将其用作语音信号的非线性预测器,分析了DWNN的函数学习能力和对高维函数学习的优越性。由于反馈单元的内部记忆能力,DWNN具有对长时相关的预测能力并能在一定程度上克服小波神经网络的“维数灾难”现象。在对语音信号的预测中,动态小波神经网络预测器的预测性能很好,虽然其预测阶数很低(仅为3),试验结果表明:DWNN预测器较RNN,RBF更适合于语音信号的非线性预测,而且其计算复杂度相对较低。更多还原

【Abstract】 In this paper, we present a dynamic wavelet neural network (DWNN) with a feedback and use it for nonlinear predictor to speech signal. Also, Its ability to learn function and superiority in appropriating the high dimensional function are analyzed. Because of the inner memory of feedback unit, dynamic wavelet neural network performs well in learning long-tern dependences and can overcome the “curse of dimension” problem that often occurs in wavelet networks to some degree. The nonlinear predictor based on DWNN has high prediction though predictive order is low (only 3) when used to speech signals. The test results indicate that, DWNN seems to be more promising than RNN and RBF network for speech prediction. Additionally, its complexity is low comparatively.更多还原