A Plug and Play Spoken Dialogue Interface for Smart Environments

In this paper we present a plug and play dialogue system for smart environments. The environment description and its state are stored on a domain ontology. This ontology is formed by entities that represent real world contextual information and abstract concepts. This information is complemented with linguistic parts that allow to automatically create a spoken interface for the environment. The spoken interface is based on multiple dialogues, related to every ontology entity with linguistic information. Firstly, the dialogue system creates appropriate grammars for the dialogues. Secondly, it creates the dialogue parts, employing a tree structure. Grammars support the recognition process and the dialogue tree supports the interpretation and generation processes. The system is being tested with a prototype formed by a living room. Users may interact with and modify the physical state of this living room environment by means of the spoken dialogue interface.