Automatic Speech Recognition Using Segments