Speech Enhancement Using Inter-Frame Dependence in DCT Domain

This paper investigates the time correlation between successive speech components and presents a novel speech enhancement algorithm taking into account the inter-frame dependence information in the discrete cosine transform (DCT) domain. The algorithm does not rely on any statistic model and can efficiently attain the optimal estimation of clean speech components from successive noisy speech components using minimum mean square error (MMSE) estimation. Moreover, it overcomes the disadvantage of independent assumption in conventional method for speech components estimation. Simulation results demonstrate that the proposed algorithm possesses good performance both in objective and subjective tests with different kinds of noise