<?xml version="1.0" standalone="yes"?>
<Paper uid="P04-1009">
  <Title>Developing A Flexible Spoken Dialog System Using Simulation</Title>
  <Section position="9" start_page="0" end_page="0" type="concl">
    <SectionTitle>
7 Conclusions and Future Work
</SectionTitle>
    <Paragraph position="0"> The use of a simulator has greatly facilitated the development of our dialog system, with the availability of thousands of artificial dialogs. Even relatively restricted synthetic dialogs have already accelerated development. In the next phase, real user data collection will be conducted, along with full-scale evaluation. We plan to compare the efficacy of our language models built from simulated data with those trained from real user data.</Paragraph>
    <Paragraph position="1"> Future research will address issues of graceful recovery from recognition error. We believe that the framework of using simulated dialogs possibly with synthesized speech input augmented with controlled levels of additive noise can be an effective way to develop and evaluate error recovery strategies.</Paragraph>
    <Paragraph position="2"> Current methods for simulating dialogs are quite rudimentary. The text only produces certain variants that have been observed but does not respect corpus statistics, nor, in the case of synthetic speech, do they account for spontaneous speech phenomena.</Paragraph>
    <Paragraph position="3"> Improved simulations could use a set of indexed real speech waveforms invoked by the core simulator to create more realistic input.</Paragraph>
    <Paragraph position="4"> The main functionalities in the simulator software are now customizable from an external file.</Paragraph>
    <Paragraph position="5"> The simulator is domain independent and can be tailored for development of similar spoken dialog systems for browsing and navigating large databases.</Paragraph>
    <Paragraph position="6"> However further work is needed to incorporate greater configurability to the dialog flow. Increased flexibility for customizing the model of the dialog is needed to enable the software to be applied to the development of other kinds of dialog systems.</Paragraph>
  </Section>
class="xml-element"></Paper>