Segmenting discrete data representing continuous speech input | IBM Journals & Magazine | IEEE Xplore