一种低信噪比环境下的基音检测方法

摘要

噪声是低信噪比环境下影响基音检测准确率的主要因素之一,为此提出一种基于形态学滤波和小波变换相结合的基音检测方法.该方法首先用形态学滤波器滤除噪声,突出基音.然后在小波域利用Teager能量算子区分清、浊音,通过浊音小波系数模的极大值提取基音.实验结果表明,在信噪比较小时该方法也能准确地检测出语音信号的基音,与传统的基音检测方法相比,该方法有较强的抗噪性.%Noise is one of the main factors that affects accuracy of pitch detection in the low SNR environment. A new method of pitch detection was proposed, which consisted of morphological filtering and wavelet transform. In the method,an algorithm based on the morphological filter was performed first to remove the noise and highlight pitch,then in the wavelet domain,to distinguish unvoiced from voiced with the teager energy operator (TEO) and to extract the pitch of the speech signal with the maximum modulus of voiced wavelet coefficients. The experimental results show that the method can accurately detect the pitch of the speech signal in low SNR and has strong noise immunity compared with other traditional methods of pitch detection.