<?xml version="1.0" standalone="yes"?>
<Paper uid="H92-1066">
  <Title>RAPID MATCH TRAINING FOR LARGE VOCABULARIES</Title>
  <Section position="7" start_page="330" end_page="331" type="concl">
    <SectionTitle>
5. FUTURE PLANS
</SectionTitle>
    <Paragraph position="0"> The work described above is only the beginning of a long-term project to improve the performance of Dragon's rapid match algorithm for continuous speech recognition. Most immediately, we plan to work at tuning the many parameters involved in rapid match training. In the results cited above, we deliberately chose parameter settings as close as possible to those used in our original training routine, using, for example, three smooth frames of length four with each of the four acoustic frames given equal weight. But there is no reason to believe that these values optimize performance. In our isolated word recognizer, in contrast, rapid match uses five smooth frames</Paragraph>
  </Section>
class="xml-element"></Paper>