汉语人机接口中神经网络模型的研究

INVESTIGATION ON NEURAL NETWORKS MODEL IN MAN--TO--COMPUTER SPEECH INTERFACES

在简要地介绍了人工神经网络(ANN)基本特性的基础上,着重介绍了多层神经网络的一种学习算法——反向传播法在语音合成中的应用,提出了一种从文本到声音的转换即完成汉语语音合成的神经网络模型.初步的实验结果是令人满意的.

Basic properties of artificial neural networks are introduced in this paper. Applications of multi-layer neuralnetworks learning algorithm, or back-propagation network, in Chinese speech synthesis are investigated. And akind of neural network model that performs text-to-speech conversion is put forward. Our preliminary experi-ment result is quite satisfactory.