<?xml version="1.0" standalone="yes"?>
<Paper uid="N06-1053">
  <Title>Towards Spoken-Document Retrieval for the Internet: Lattice Indexing For Large-Scale Web-Search Architectures</Title>
  <Section position="2" start_page="0" end_page="415" type="intro">
    <SectionTitle>
1 Introduction
</SectionTitle>
    <Paragraph position="0"> Search engines have become the essential tool for finding and accessing information on the Internet. The recent runaway success of podcasting has created a need for similar search capabilities to find audio on the web.</Paragraph>
    <Paragraph position="1"> As more news video clips and even TV shows are offered for on-demand viewing, and educational institutions like MIT making lectures available online, a need for audio search arises as well, because the most informative part of many videos is its dialogue.</Paragraph>
    <Paragraph position="2"> There is still a significant gap between current web audio/video search engines and the relatively mature text search engines, as most of today's audio/video search engines rely on the surrounding text and metadata of an audio or video file, while ignoring the actual audio content. This paper is concerned with technologies for searching the audio content itself, in particular how to represent the speech content in the index.</Paragraph>
    <Paragraph position="3"> Several approaches have been reported in the literature for indexing spoken words in audio recordings. The TREC (Text REtrieval Conference) Spoken-Document Retrieval (SDR) track has fostered research on audioretrieval of broadcast-news clips. Most TREC benchmarking systems use broadcast-news recognizers to generate approximate transcripts, and apply text-based information retrieval to these. They achieve retrieval accuracy similar to using human reference transcripts, and ad-hoc retrieval for broadcast news is considered a &amp;quot;solved problem&amp;quot; (Garofolo, 2000). Noteworthy are the rather low word-error rates (20%) in the TREC evaluations, and that recognition errors did not lead to catastrophic failures due to redundancy of news segments and queries. However, in our scenario, unpredictable, highly variable acoustic conditions, non-native and accented speaker, informal talking style, and unlimited-domain language cause word-error rates to be much higher (40-60%). Directly searching such inaccurate speech recognition transcripts suffers from a poor recall.</Paragraph>
    <Paragraph position="4"> A successful way for dealing with high word error rates is the use of recognition alternates (lattices) (Saraclar, 2004; Yu, 2004; Chelba, 2005). For example, (Yu, 2004) reports a 50% improvement of FOM (Figure Of Merit) for a word-spotting task in voice-mails, and (Yu, HLT2005) adopted the approach for searching personal audio collections, using a hybrid word/phoneme lattice search.</Paragraph>
    <Paragraph position="5"> Web-search engines are complex systems involving substantial investments. For extending web search to audio search, the key problem is to find a (approximate)  representation of lattices that can be implemented in a state-of-the-art web-search engine with as little changes as possible to code and index store and without affecting its general architecture and operating characteristics.</Paragraph>
    <Paragraph position="6"> Prior work includes (Saraclar, 2004), which proposed a direct inversion of raw lattices from the speech recognizer. No information is lost, and accuracy is the same as for directly searching the lattice. However, raw lattices contain a large number of similar entries for the same spoken word, conditioned on language-model (LM) state and phonetic cross-word context, leading to inefficient usage of storage space.</Paragraph>
    <Paragraph position="7"> (Chelba, 2005) proposed a posterior-probability based approximate representation in which word hypotheses are merged w.r.t. word position, which is treated as a hidden variable. It easily integrates with text search engines, as the resulting index resembles a normal text index in most aspects. However, it trades redundancy w.r.t. LM state and context for uncertainty w.r.t. word position, and only achieves a small reduction of index entries. Also, time information for individual hypotheses is lost, which we consider important for navigation and previewing.</Paragraph>
    <Paragraph position="8"> (Mangu, 2000) presented a method to align a speech lattice with its top-1 transcription, creating so-called &amp;quot;confusion networks&amp;quot; or &amp;quot;sausages.&amp;quot; Sausages are a parsimonious approximation of lattices, but due to the presence of null links, they do not lend themselves naturally for matching phrases. Nevertheless, the method was a key inspiration for the present paper.</Paragraph>
    <Paragraph position="9"> This paper is organized as follows. The next section states the requirements for our indexing method and describes the overall system architecture. Section 3 introduces our method, and Section 4 the results. Section 5 briefly describes a real prototype built using the approach.</Paragraph>
  </Section>
class="xml-element"></Paper>