Comment on “Blind source separation based on endpoint estimation with applications to the MLSP 2006 data competition”