Text this: Method of speakers segmentation based on pre-segmentation