In this demo, the target (a loudspeaker playing three male and three female utterances each of length 4 s) moves continuously from position 1 to 16. A video of the recording is here. There are an interfering woman speaker at the fixed position JF and a moving man speaker along the path JM. After six seconds the interfering speakers are interchanged so the next six seconds is the man in position JF and the woman moves along the path JM. CHiME background noise is added (30 dB SNR) and a Gaussian white noise (50 dB SNR). To extract the noise, the semi-blind method utilizes target-cancellation filters that are known for positions 1 to 16 as a priori information.
Signals obtained via our semi-blind method 
Signals obtained via a semi-blind method derived from 
-  Z. Koldovský, J. Málek, P. Tichavský, and F. Nesta, “Semi-blind Noise Extraction Using Partially Known Position of the Target Source”, IEEE Trans. on Speech, Audio and Language Processing, vol. 21, no. 10, pp. 2029-2041, Oct. 2013. (here)
-  F. Nesta and M. Omologo, “Convolutive Underdetermined Source Separation through Weighted Interleaved ICA and Spatio-temporal Source Correlation”, Proc. of The 10th International Conference on Latent Variable Analysis and Source Separation (LVA/ICA 2012), pp. 222-230, Tel-Aviv, Israel, March 12-15, 2012.