Speech interfaces are becoming more and more popular as a means to interact with virtual environments but the development and integration of these interfaces is usually still ad-hoc, especially the speech grammar creation of the speech interface is a process commonly performed by hand. In this paper, we introduce an approach to automatically generate a speech grammar which is generated using semantic information. The semantic information is represented through ontologies and gathered from the conceptual modelling phase of the virtual environment application. The utterances of the user will be resolved using queries onto these ontologies such that the meaning of the utterance can be resolved. For validation purposes we augmented a city park designer with our approach. Informal tests validate our approach, because they reveal that users mainly use words represented in the semantic data, and therefore also words which are incorporated in the automatically generated speech grammar.

Part of the research at EDM is funded by ERDF (European Regional Development Fund), the Flemish Government and the Flemish Interdisciplinary institute for BroadBand technology (IBBT). The VR-DeMo project (IWT 030248) is directly funded by the IWT, a Flemish subsidy organization. The authors would like to thank Geert Vanderhulst and the VUB-WISE partner for their help and advice.