为了在不同噪声、信噪比下为基音检测算法提供更能准确反映基音周期实际变化的输入语音,本文将信号分解思想引入基音检测前端处理中。
In order to provide an accurate-pitch-cycle speech for pith detection algorithm with varied noise and SNR, we use signal decomposition theory in pre-processing of pitch detection.
但研究发现在极低信噪比,由于观测信号的样本协方差矩阵具有奇异性,这使得ICA去噪算法中的白化处理步骤无法进行。
But in the very low SNR circumstance, because of the covariance matrix of the observed signals being singularity, the ICA denoising method can not be used.
应用推荐