System of an Automatic Speech Recognition and Speech Understanding LINGVO/LASER

Authors: Václav Matoušek, Kamil Ekštein, Tomáš Pavelka

Abstract:

The presented paper introduces a scalable modular prototyping concept and methodology framework for rapid development of domain-oriented information retrieval dialogue systems being developed in co-operation of the above mentioned department and the Department of Informatics of the Faculty of Art IV of the University of Regensburg. The system consists of two stand-alone aggregates of software modules and knowledge bases – speech recognition and synthesis software (LASER) and a generic task-independent scalable dialogue system (LINGVO). The article presents the architecture, design, and philosophy of the framework in details and discusses its benefits experienced in development of an experimental dialogue information system (DIS).