<?xml version="1.0" standalone="yes"?>
<Paper uid="A00-1012">
  <Title>Preceding word</Title>
  <Section position="3" start_page="0" end_page="0" type="intro">
    <SectionTitle>
1 Speech recognition systems are often evaluated in terms
</SectionTitle>
    <Paragraph position="0"> of word error rate (WER), the percentage of tokens which are wrongly transcribed. For large vocabulary tasks and speaker-independent systems, WER varies between 7% and 50%, depending upon the quality of the recording being recognised.</Paragraph>
    <Paragraph position="1"> See, e.g., Cole (1996).</Paragraph>
  </Section>
class="xml-element"></Paper>