论文部分内容阅读
提出了一种可用于说话人识别的神经阵列网络，它以仅完成两类模式区分的小型网络作为子网络，再将单个子网络组合成阵列形式来完成多类模式的区分。文中给出了阵列网络的构成及搜索算法，并使用径向基函数（ＲＢＦ）阵列网络进行了文本无关的说话人识别的研究。实验显示，对 ２０名说话人，用 ５秒语音训练， ２秒语音识别时，该方法可达到 ９８％的正确识别率。
A novel neural network which can be used for speaker recognition is proposed. It takes a small network that only completes the two types of modes as a subnetwork, and then combines individual subnetworks into an array to accomplish the distinction between multiple classes. In this paper, the structure of the array network and its search algorithm are given, and the text-independent speaker recognition is studied by using radial basis function (RBF) array network. Experiments show that this method can achieve a correct recognition rate of 98% when using 20 seconds of voice training with 5 seconds of voice training and 2 seconds of voice recognition.