On using a priori segmentation of the speech signal in an N-best solutions post-processing | IEEE Conference Publication | IEEE Xplore