Figure 2From: An iterative model-based approach to cochannel speech separation SNR gains of target speaker at different input SNR conditions with beam width varying from 1 to 256. Back to article page