摘要
本文以BP神经网络算法和语音端点检测为基础,先分析了神经网络发展的历史以及语音端点检测的背景,然后着重分析了BP神经网络的算法训练原理,分析了MFCC参数的提取过程,最后以MFCC参数为训练特征向量,结合BP神经网络进行训练,最后得出结论该算法检测识别率较高,整体效果较好。
Based on BP neural network algorithm and speech endpoint detection,this paper first analyzes the history of neural network development and the background of speech endpoint detection.Then it analyzes the algorithm training principle of BP neural network and analyzes the extraction process of MFCC parameters.The MFCC parameters are training eigenvectors,combined with BP neural network for training.Finally,it is concluded that the algorithm has higher detection and recognition rate and better overall effect.
作者
李震
Li Zhen(Xi’an Technological University,Xi’an,710021)
出处
《数字通信世界》
2019年第3期15-15,38,共2页
Digital Communication World