摘要
提出了一种基于经验模式分解的平均幅度差函数基音检测改进算法.该算法首先求取每帧浊音语音的AMDF;而后利用EMD对AMDF进行分解,得到有限个本征模式函数和残余分量,利用所有的IMF重构组成新的EMDAMDF;最后利用EMDAMDF检测出该语音帧的基音.仿真结果表明,由于能够有效地去除传统AMDF的均值下降趋势,该算法的性能明显优于传统的AMDF和CAMDF.
This paper presents a modified Average Magnitude Difference Function（ AMDF）algorithm based on Empirical Mode Decomposition（EMD） for pitch detection. Firstly, AMDF of the voiced speech frame is computed. Then AMDF is decomposed into a residue component and a finite set of band-limited signals termed as Intrinsic Mode Functions（IMFs） by using EMD. A modified AMDF called EMDAMDF is reconstructed by all IMFs. Finally ,the true pitch can be detected from EMDAMDF. The simulated results show that the performance of the proposed algorithm is much better than that of the conventional AMDF and CAMDF as the falling trend is eliminated effectively.
出处
《南京师范大学学报（工程技术版）》
CAS
2013年第1期62-67,共6页
Journal of Nanjing Normal University(Engineering and Technology Edition)
基金
江苏省自然科学基金(BK2010546)
关键词
基音
平均幅度差函数
经验模式分解
本征模式函数
pitch, average magnitude difference function （ AMDF）, empirical mode decomposition （ EMD）, intrinsic modefunctions （IMFs）
