Speech enhancement using transfer function ratio beamformer and matched filter array

Human machine interface for speech processing often suffers from interferences which contain nonstationary signal, stationary signal and reverberation. In this paper, we consider two speech sources in the reverberant and noisy environment and we develop a speech enhancement approach to extract desired speech signal from the corrupt observations. The proposed method uses transfer function ratio beamformer, H_infin adaptive filter algorithm and matched filter array to perform speech enhancement as well as dereverberation at two stages. The desired speech is extracted at the first stage and the extracted speech which contains reverberation is de-reverberated at the second stage.