Adapting multimodal dialog for the elderly

In this paper, we outline the design of a multimodal interface for a mobile pedestrian navigation system developed within the project COLLATE. The aim of the interface is to adapt to different resource limitations of the user. It takes into account the cognitive load of the user as well as the age. We present an approach on how special acoustic models for elderly speakers can improve speech recognition quality and at the same time provide an information source for user modeling. Three different presentation strategies are presented: unimodal (speech only, graphics only), redundant (speech and graphics providing the same information), and concurrent (minimally overlapped speech and graphics).