摘要
自相关函数法是基音周期提取中一种简单而有效的检测算法,针对传统的自相关函数法在低信噪比环境下鲁棒性较差的问题,提出一种改进的基音周期提取算法。该算法通过谱减法对语音信号降噪,随后进行端点检测,并提取元音的主体,在自相关函数的基础上进行改进,对元音主体及过渡区间进行基音检测。实验结果表明,该算法具有较高的准确率,与传统自相关检测算法相比,鲁棒性明显提高。
Autocorrelation function(ACF)method is a simple and effective detection algorithm in pitch periodic extraction.Aiming at the poor robustness of the traditional ACF method in a low SNR environment,an improved pitch periodic extraction algorithm is proposed.In this algorithm,the spectral subtraction method is used to reduce noise in speech signals,endpoint detection is performed,main vowels are extracted and improved on the basis of the ACF,and pitch detection is performed for mainvowels and the transition zone.The experimental results show that the algorithm has a higher accuracy rate,and the robustnessis improved significantly in comparison with the traditional autocorrelation detection algorithm.
作者
韩芳
王学春
靳宗信
HAN Fang;WANG Xuechun;JIN Zongxin(School of Information Engineering,Huanghe Science & Technology College,Zhengzhou 450063,China)
出处
《现代电子技术》
北大核心
2017年第19期71-74,78,共5页
Modern Electronics Technique
基金
国家自然科学基金项目(61101232
51472102)
郑州市科技局科技发展计划项目(20140663)
郑州市嵌入式系统应用技术重点实验室建设项目(121PYFZX177)
关键词
基音周期
自相关函数
谱减法
基音检测
语音处理
pitch period
autocorrelation function
spectral subtraction
pitch detection
speech processing
