Towards Spoken-Document Retrieval for the Internet: Lattice Indexing For Large-Scale Web-Search Architectures

Large-scale web-search engines are generally designed for linear text. The linear text representation is suboptimal for audio search, where accuracy can be significantly improved if the search includes alternate recognition candidates, commonly represented as word lattices.This paper proposes a method for indexing word lattices that is suitable for large-scale web-search engines, requiring only limited code changes.The proposed method, called Time-based Merging for Indexing (TMI), first converts the word lattice to a posterior-probability representation and then merges word hypotheses with similar time boundaries to reduce the index size. Four alternative approximations are presented, which differ in index size and the strictness of the phrase-matching constraints.Results are presented for three types of typical web audio content, podcasts, video clips, and online lectures, for phrase spotting and relevance ranking. Using TMI indexes that are only five times larger than corresponding linear-text indexes, phrase spotting was improved over searching top-1 transcripts by 25-35%, and relevance ranking by 14%, at only a small loss compared to unindexed lattice search.