<?xml version="1.0" standalone="yes"?>
<Paper uid="H89-2018">
  <Title>THE COLLECTION AND PRELIMINARY ANALYSIS OF A SPONTANEOUS SPEECH DATABASE*</Title>
  <Section position="11" start_page="132" end_page="132" type="concl">
    <SectionTitle>
SUMMARY
</SectionTitle>
    <Paragraph position="0"> This paper documents our initial effort in developing a spontaneous speech database, and reports some preliminary analyses of the collected data. We found the process of data collection to be relatively straightforward, and we believe it will be fairly easy to collect more data at a later stage, after VOYAGER's capabilities have improved. In fact, we believe that incremental data collection done this way can be quite effective for development of spoken language systems.</Paragraph>
    <Paragraph position="1"> Our preliminary analysis of these data has already indicated some significant differences between spontaneous and read speech. We are also beginning to use the database to train and evaluate the VOYAGER system, both at the acoustic-phonetic and the linguistic levels. The results of the preliminary evaluation are described in a companion paper \[6\].</Paragraph>
  </Section>
class="xml-element"></Paper>