基于形态学滤波和HHT的基音检测方法

发布时间：2018-01-05 23:29

本文关键词：基于形态学滤波和HHT的基音检测方法　出处：《铁道学报》2014年07期 　论文类型：期刊论文

【摘要】：提出一种基于形态学滤波和Hilbert-Huang变换HHT(Hilbert-Huang Transform)的基音检测方法,该方法先对带噪语音信号进行形态学滤波,滤除噪声,突出基音,滤波后的信号进行Hilbert-Huang变换得到信号的Hilbert谱,由Hilbert谱计算瞬时能量及其导数,通过瞬时能量的突变定位声门开闭的时刻来精确地跟踪基音的变化。与传统的基音检测方法向比较,该方法真实地描述了语音信号非平稳、非线性的特性,对语音信号的分析是一种自适应的过程,实验结果表明,该方法有较强的抗噪性,在低信噪比环境下仍能够精确地提取出语音信号的基音。
[Abstract]:A pitch detection method based on morphological filtering and Hilbert-Huang transform HHT(Hilbert-Huang transform is proposed. This method firstly carries on the morphological filtering to the noisy speech signal, filters the noise, highlights the pitch, the filtered signal carries on the Hilbert-Huang transformation to obtain the signal Hilbert spectrum. The instantaneous energy and its derivative are calculated by Hilbert spectrum, and the change of pitch is accurately tracked by the sudden change of instantaneous energy to locate the opening and closing moment of glottis, which is compared with the traditional pitch detection method. The method describes the non-stationary and nonlinear characteristics of speech signal, and the analysis of speech signal is an adaptive process. The experimental results show that the method has strong anti-noise. In low SNR environment, the pitch of speech signal can be extracted accurately.
【作者单位】：兰州交通大学电子与信息工程学院;
【基金】：国家自然科学基金(61261029,61202314)
【分类号】：TN912.3
【正文快照】： 语音信号是一种时变的非平稳过程,加上声道共振峰和噪声的影响,准确的基音检测十分困难。目前基音检测的方法主要有自相关法、逆滤波法、平均幅度差函数法、倒谱法、并行处理法、谱相减法和过零数法等[1-3],属于线性处理的方法,并不考虑声门闭合的准确时刻和瞬时性,而是计算一

【相似文献】