Novel Laplacian Factor Estimation Algorithm for Speech Enhancement

摘要

Based on the property of generalized Gaussian distribution model and its shape parameter, a novel approach for Laplacian factor estimation is presented, which indirectly attains the estimation of Laplacian factor using its relation with the variance of clean speech components. As for the estimation for speech components variance, a new algorithm is used by making use of the noisy speech components and clean speech variance in previous frame to compute the current frame's speech variance. By combining the given two approaches, the estimated Laplacian factor can not be affected by noise components energy and the obtained result keeps accurate. Simulation results demonstrate that the proposed algorithm possesses good performance under different kinds of noise.