抄録
'’KanNon System'' is the software of a sound visualization system, which is used to help in communication the people who have a hearing difficulty. Having the input as a voice signal, the output is done by showing string of characters, pitch of voice and sound spectrogram. The system applies a method of statistical pattern recognition for phoneme. To show character strings, accurate speech recognition is required. In this research, a method of statistically multiplexing pattern is presented to get the accurate recognition.
