基于神经网络的与文本相关说话人辨认系统

说话人识别是语音识别的一种特殊方式，对于各种领域的身份认证具有得天独厚的优势。以线性预测系数（Linear Predielion Coefficients，LPC）作为特征参数，采用多层感知器神经网络和BP算法建立了一个与文本相关的说话人辨认系统。实验结果表明，这种神经网络系统在说话人辨认中是有效的。

A Neural Network Based Text-dependent Speaker Identification System

Speaker recognition is a special mode of speech recognition, which has many exclusive advantages in identity authentification. We constract a neural network based text-dependent speaker identification system, in which Linear Prediction Coeffcients （LPC） are extracted from speech signals and serve as feature vectors. These vectors are fed into a muhilayer perceptron （MLP） neural network with backpropagation learning algorithm for its training and speaker identification. Experimental results indicate that this system is effective for speaker identification.