【摘要】 在双频点频域卷积混合盲分离算法的基础上,提出一种单频点频域卷积混合盲分离算法.通过恰当地选取频域盲分离中短时傅里叶变换(STFT)的参数,可仅依靠一个频率片上的分离信号来完成双频点算法.相对于传统频域卷积混合盲分离方法,该方法不受各频率片顺序不确定性不统一及比例不确定性不统一的影响.仿真实验证明了该算法的有效性.更多还原

【Abstract】 In this paper,a novel frequency domain separation algorithm using only one frequency bin for the convolutive mixtures of sources is proposed.The method is a variation of the double-frequency-bin based separation for convolutive mixtures in which the parameters of short time Fourier transform should be selected appropriately.Since only one frequency bin is used,the proposed method involves neither inconsistency of scale indeterminacy nor permutation indeterminacy between different frequency bins in traditional frequency domain separation of convolutive mixtures.Simulation experiments show validity of the proposed single-frequency-bin based separation algorithm for convolutive mixtures.更多还原