<?xml version="1.0" standalone="yes"?>
<Paper uid="H90-1036">
  <Title>A Rapid Match Algorithm for Continuous Speech Recognition</Title>
  <Section position="4" start_page="171" end_page="171" type="concl">
    <SectionTitle>
4. Conclusions
</SectionTitle>
    <Paragraph position="0"> The rapid match procedure that we have described here appears to be a very promising method. On an approximately 1000 word continuous speech recognition task--a task that is not artificial, although another 1500 words would need to be added to give adequate coverage of real mammography reports in actual practice--it has enabled us to obtain near real time performance on a 33 megahertz 80486 processor.</Paragraph>
    <Paragraph position="1"> We have done some preliminary work on adapting to new speakers based on several hundred mammography sentences, and we are optimistic that our rapid match models will adapt successfully, as they have inside of DragonDictate. A future paper will discuss the general question of the training and adaptation of these models.</Paragraph>
  </Section>
class="xml-element"></Paper>