A New and Effective Technique for Blind RT Estimation Via Laplacian Decay Model

Abstract

In this research a statistical model based on Laplacian distribution is adopted for modeling the reverberant signal energy decay, and then the maximum likelihood (ML) estimator is used to estimate its reverberation time (RT). Based on the study of existing technique for RT estimation, which uses Gaussian random process for modeling the reverberation decay; a decay model based on Laplacian distribution is presented in this work which includes an efficient method for detecting free decays from the reverberant signal and then using these decays RT is estimated with the help of ML estimator. Our observation about the closeness of temporal decay distribution pattern of the reverberant hand clap signal to the Laplacian distribution was the motivation behind this research. Experimental results show that the proposed algorithm is more efficient and provides accuracy in RT estimation.