An Efficient Multistage Approach for Blind Source Separation of Noisy Convolutive Speech Mixture

This paper proposes a novel efficient multistage algorithm to extract source speech signals from noisy convolutive mixture. The proposed approach comprises two stages named Blind Source Separation (BSS) and de-noising. A hybrid prior model separates the reverberant mixture in BSS stage. Moreover, we low- high-energy components by generalized multivariate Gaussian super-Gaussian models, respectively. We use Minimum Mean Square Error (MMSE) reduce noise signal de-noising Furthermore, models investigate performance gain. In first model, is separated observed stage, followed suppression of estimated module. second approach, reduced using MMSE filtering technique received at separation de-noised evaluate scheme terms signal-to-distortion ratio (SDR) with respect other well-known methods. results show superior over state-of-the-art