A variable step-size multichannel equalization algorithm exploiting sparseness measure for room acoustics

Non-adaptive multichannel equalization (MCEQ) algorithms based on multiple input/output inverse theorem (MINT) is computationally expensive as MINT involves the inversion of a convolution matrix with dimension that is proportional to the length of the acoustic impulse responses. To address this, we propose a MINT-based algorithm that estimates inverse filters by minimizing a cost function iteratively. To further enhance the convergence rate, we formulate an algorithm that employs an adaptive step-size that is derived as a function of the sparseness measure. The proposed algorithm is then applied to existing MINT-based equalization algorithms such as A-MINT and the currently proposed MCEQ-based algorithms.