Sound source separation based on non-negative tensor factorization incorporating spatial cue as prior knowledge | IEEE Conference Publication | IEEE Xplore