<?xml version="1.0" standalone="yes"?>
<Paper uid="P01-1002">
  <Title>Processing Broadcast Audio for Information Access</Title>
  <Section position="9" start_page="0" end_page="0" type="concl">
    <SectionTitle>
9 Conclusions
</SectionTitle>
    <Paragraph position="0"> This paper has described some of the ongoing research activites at LIMSI in automatic transcription and indexation of broadcast data. Much of this research, which is at the forefront of todays technology, is carried out with partners with real needs for advanced audio processing technologies. null Automatic speech recognition is a key technology for audio and video indexing. Most of the linguistic information is encoded in the audio channel of video data, which once transcribed can be accessed using text-based tools. This is in contrast to the image data for which no common description language is widely adpoted. A variety of near-term applications are possible such as audio data mining, selective dissemination of information (News-on-Demand), media monitoring, content-based audio and video retrieval. It appears that with word error rates on the order of 20%, comparable IR results to those obtained on text data can be achieved. Even with higher word error rates obtained by running a faster transcription system or by transcribing compressed audio data (Barras et al., 2000; J.M. Van Thong et al., 2000) (such as that can be loaded over the Internet), the IR performance remains quite good.</Paragraph>
  </Section>
class="xml-element"></Paper>