In this paper we describe a two-channel speech enhancement system. Two noisy versions of a speech signal are picked up with closely spaced microphones. The system systematically exploits temporal and spatial characteristics of the sound field and fully operates in the frequency domain. The system is based on the spectral subtraction technique and adaptive beamforming. It may be used to improve speech quality for hands-free telephone as well as to preprocess speech for speech operated systems.
Bibliographic reference. Linhard, Klaus (1995): "Speech enhancement using two versions of the noisy speech signal", In EUROSPEECH-1995, 2005-2008.