摘要
语音端点检测计算的准确性,是影响语音通信质量和通信效率的最主要因素,通过加强对语音端点检测算法的研究,可以有效提高语音端点检测计算的精准性,对推动我国语音识别技术的发展和进步意义重大。几种传统形式的语音端点检测算法都存在一定的缺陷,文章提出了一种基于小波神经网络的语音端点检测算法,并对其系统结构进行了分析,以语音与噪声的频域差异及统计自相似性为理论依据,通过提取语音特征量构建相应的算法模型,对仿真实验结果进行分析,证明该算法可以有效提高语音端点检测计算的准确性。
The accuracy of speech endpoint detection calculation is the main factors that affect the quality of speech communication and communication efficiency. By strengthening the study on speech endpoint detection algorithm,the precision of calculation of speech endpoint detection is effectively improved. It is of great significance to promote the development and progress of our country＇s speech recognition technology. Speech endpoint detection algorithm of several traditional forms have some defects,this paper proposes a speech endpoint detection algorithm based on wavelet neural network,and the system structure is analyzed in frequency domain between speech and noise and statistical self similarity theory,the speech feature extraction algorithm to construct the corresponding model volume and the analysis of simulation results show that the algorithm can effectively improve the accuracy of speech endpoint detection calculation.
作者
孙护军
SUN Hujun(School of Electronic Engineering,Xi＇an Aeronautical University,Xi＇an 710077)
出处
《计算机与数字工程》
2018年第9期1717-1720,1818,共5页
Computer & Digital Engineering
关键词
小波神经网络
语音端点检测
算法模型
仿真实验
wavelet neural network
speech endpoint detection
algorithm model
simulation experiment
