基于神经网络的非特定人数字符汉语语音识别

摘要

In this paper, we present a kind of speech recognition system using a high-order Pi-Sigma neural network. The system is designed to recognize the ten digital characters spoken in Chinese. With the help of TMS320C25 high-speed digital signal processor, we prepared voice samples collected from different speakers and achieve a recognition rate of 98% by this system. We introduce the modular construct...