Abstract
We develop the real-time speech visualization system called “KanNon”[1, 2] which supports speech communication of deaf people. The KanNon system presents several information of the speech such as loudness, pitch, sound spectrogram and characters by speech recognition system in real-time. In the present system, we are adapting a word unit speech recognition system using large-scale dictionary. However the KanNon system is required quick and simple display of speech contents for smooth communication. For this purpose, we apply phonemic speech recognition system for Japanese 5 vowels using “Time-Delay Neural Network (TDNN)”. Further, we developed speech detection, voiced/unvoiced (v/uv) detection and change detection algorithms in the KanNon system. Finally, we show experimental results using real speech data.
