Making Speech Look Like Text in the Regulus Development Environment

Abstract

We present an overview of the development environment for Regulus, an Open Source platform for construction of grammar-based speech-enabled systems, focussing on recent work whose goal has been to introduce uniformity between text and speech views of Regulus-based applications. We argue the advantages of being able to switch quickly between text and speech modalities in interactive and offline testing, and describe how the new functionalities enable rapid prototyping of spoken dialogue systems and speech translators.