<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1622">
  <Title>Adding extra input/output modalities to a spoken dialogue system</Title>
  <Section position="4" start_page="2" end_page="2" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> For a long time, speech has been the only modality for input and output in telephone-based information systems. Speech is often considered to be the most natural form of input for such systems, since people have always used speech as the primary means of communication. Moreover, to use a speech-only system a simple telephone suffices and no additional devices are required. Obviously, in situations where both hands and eyes are busy, speech is definitely preferable over other modalities like pen/mouse.</Paragraph>
    <Paragraph position="1"> However, speech-only interfaces have also shown a number of shortcomings that result in less effective and less efficient dialogues.</Paragraph>
    <Paragraph position="2"> The aim of the research described in this paper is to assess the extent to which multimodal input/output can help to improve effectiveness, efficiency and user satisfaction of information systems in comparison with unimodal systems.</Paragraph>
    <Paragraph position="3"> This paper describes how, within the framework of the MATIS  (Multimodal Access to Transaction and Information Services) project we developed a prototype of a multimodal railway information system by extending a speech-only version in such a way that it supports screen output and point-and-click actions of the user as input. This system is a typical example of a simple application that can be implemented using a slot-filling paradigm and may stand model for various other form filling applications.</Paragraph>
    <Paragraph position="4"> First, a number of problems are described that arise in speech-only interfaces. Then we briefly describe the architecture of the speech-only  http://www.ipo.tue.nl/projects/matis/ railway information system. Next, we describe in more detail how we added multimodality to this version of the system and explain why we think this may help to solve the shortcomings of speech-only systems. We conclude this paper by discussing several open issues that we intend to solve by means of user tests with the multimodal system.</Paragraph>
  </Section>
class="xml-element"></Paper>