W03-0603 audio was segmented using our speech segmentation algorithm based on pause structure ( Yoshida
hide detail