A "Bunsetsu" speech recognition system using a top-down processor controlled by bottom-up information.

Akio Ando, Kazuhiko Ozeki

Browse the full ICASSP paper archive.