摘要
逆滤波法估计声门波含有较多的谐波成分,为了减小这种误差,提出一种复倒谱相位分解的声门波估计方法。首先估计嗓音信号的基音周期,利用复倒谱将每个基音周期内嗓音信号分解为最大相位信号与最小相位信号,经过微分运算后与声门闭合点位置结合,实现微分声门波估计。对微分声门波积分,即为所估计的声门波。采用迭代自适应逆滤波方法和本文方法提取的声门波进行参数提取。实验结果表明,本文方法提取的声门波更能反映声带振动特性,声门开相、闭相信息明确,有效减小声源谐波成分和声道干扰。在估计微分声门波最小值时表现出最佳性能,估计误差低至0.25%。89%的时域参数与标准值的误差均小于10%
Using Inverse filtering to estimate the glottal flow which contains a lot of harmonic component,in order to reduce the error,a method of complex cepstrum phase decomposition is proposed to estimate glottal flow.First,pitch of voice signal is estimated,voice signal in each pitch is decomposed into maximum phase signal and minimum phase signal by using complex cepstrum,then combining with the glottal closed point position after the differential operation,it can achieve derivative glottal flow estimation.Glottal wave is obtained by integrating derivative glottal flow.Iterative adaptive inverse filtering method and the proposed method are used to extract the parameters.Experimental results show that the glottal flow reflected more vocal fold vibration characteristic,the information of open phase and close phase of glottis is clear.Interference of vocal fold and the harmonic components of voice source is effectively reduced.The estimation of derivative glottal flow shows best performance with the minimum value,estimation error can be as low as 0.254 6%.The error between 89%of the time domain parameters and standard values is less than 10%.
作者
王琰
邵雅婷
张晓俊
吴迪
曾晓亮
顾济华
陶智
WANG Yan;SHAO Yating;ZHANG Xiaojun;WU Di;ZENG Xiaoliang;Gu Jihua;TAO Zhi(College of Physics,Optoelectronics and Energy of Soochow University Soochow,Suzhou Jiangsu 215006,China)
出处
《电子器件》
CAS
北大核心
2018年第4期1055-1060,共6页
Chinese Journal of Electron Devices
基金
国家自然科学基金重点项目(61271359)
江苏省自然科学基金青年项目(BK20140354)
关键词
声门波
复倒谱
相位分解
开相
闭相
glottal flow
complex cepstrum
phase decomposition
open phase
close phase
