Tools evaluation for speech recognition based on domain ontologies over the

Abstract

Speech recognition is a natural way of interaction between people and mobile devices. However, the inclusion of all vocabulary variants for a specific language may be tricky and expensive on processing terms; also voice training is not appreciated by most of users. Speech recognition matters have been addressed from several ways, even including artificial intelligence concepts as recent IPhone Siri system, but most of them are proprietary solutions. This paper explores other research alternatives related with the use of domain ontologies to restrict the vocabulary to the application context and assist the speech recognition process. Specifically, a first approach is given by evaluating the most important components involved on the speech recognition process based on domain ontologies, using open source tools over Android platform.