This review presents an overview of computational auditory scene analysis (CASA), as biologically inspired approaches for machine sound separation. In this review, we address human auditory system containing early auditory stage, binaural combining, cortical stage, and top-down attention. We compared the models employed for CASA, especially for early auditory and cortical stages. We emphasized on how the existing models are similar to
human auditory mechanism for sound separation. Finally, we discussed current issues and future of this task.