Single-Channel Blind Source Separation using Adaptive Mode Separation- Based Wavelet Transform and Density-Based Clustering with Sparse Reconstruction

Abstract

Abstract In this study, the signal-channel blind source separation (SCBSS) problem has been addressed using a novel approach. The approach is based on combining the adaptive mode separation-based wavelet transform with adaptive mode separation (AMSWT) and the density-based clustering with sparse reconstruction. The approach is performed in Time frequency domain and in reverberant environment. First, using the Fourier transform, the amplitude spectrum of the observed mixture signal is obtained. Then, using variational scaling and wavelet functions, the AMSWT is introduced to adaptively extract spectral intrinsic components (SIC). To obtain a better time-frequency distribution, the AMSWT is applied to each mode. Thus, the SCBSS problem is transformed into a non-underdetermined. Then, for each frequency bin; the density-based clustering, reformulated to eigenvector clustering problem, is performed to estimate the mixing matrix. Finally, the sparse reconstruction is introduced to reconstruct the estimated source. The proposed approach has been evaluated using an objective measure of separation quality. According to experimental results, the proposed approach presents a powerful method to solve the SCBSS problem, and provide better separation performances than the existing methods.