<?xml version="1.0" standalone="yes"?>
<Paper uid="W06-3711">
  <Title>IBM MASTOR SYSTEM: Multilingual Automatic Speech-to-speech Translator *</Title>
  <Section position="7" start_page="0" end_page="0" type="concl">
    <SectionTitle>
5. CONCLUSION
</SectionTitle>
    <Paragraph position="0"> We described the framework of the IBM MASTOR system, the various technologies used in building major components for languages with different levels of data resources. The technologies have shown successes in building real-time S2S systems on both laptop and small computation resource platforms for two language pairs, English-Mandarin Chinese, and English-Arabic dialect. In the latter case, we also developed approaches which lead to very rapid (in the matter of 3-4 months) development of systems using very limited language and domain resources. We are working on improving spontaneous speech recognition accuracy and more naturally integrating two translation approaches.</Paragraph>
  </Section>
class="xml-element"></Paper>