<?xml version="1.0" standalone="yes"?>
<Paper uid="H89-1017">
  <Title>THE MINDS SYSTEM: USING CONTEXT AND DIALOG TO ENHANCE SPEECH RECOGNITION</Title>
  <Section position="13" start_page="134" end_page="134" type="concl">
    <SectionTitle>
SUMMARY
</SectionTitle>
    <Paragraph position="0"> The MINDS system was built to demonstrate the feasibility of using contextual, dialog-level knowledge sources to constrain the exponential growth of a search space and hence increase speech recognition accuracy. The results of our studies using the system indicate that such knowledge sources are effective for dynamically reducing the number of word candidates a speech recognition system must consider when analyzing a speech signal. As we move towards robust, interactive problem solving environments where speech is a primary input medium, use of these knowledge sources could prove important for enhancing system performance.</Paragraph>
  </Section>
class="xml-element"></Paper>