The Evaluation of Spoken Dialog Management Models for Multimodal HCIs

The implementation of voice dialogs enables the realization of some of the aims of modern Human Computer Interaction (HCI) services more successfully and efficiently. Sadly the multimodal Lithuanian HCIs carried by the most natural form of communication-speech are still in the prototype stage and no services are provided to end user at the time of writing. This paper describes an experimental evaluation of the possibilities of using the spoken language dialogs as the main modality in modern application control. The recognition accuracy of the tree main types of spoken dialogues (dictation, keyword spotting, isolated utterances) was evaluated and user preference survey was done on proposed multimodal HCIs. The goal of this research was to gather the results by possible everyday future users not familiar with such systems.