Spoken Document Retrieval Based on Approximated Sequence Alignment

Abstract. This paper presents a new approach to spoken document information retrieval for spontaneous speech corpora. The classical approach to this problem is the use of an automatic speech recognizer (ASR) combined with standard infor-mation retrieval techniques. However, ASRs tend to produce transcripts of spon-taneous speech with significant word error rate, which is a drawback for standard retrieval techniques. To overcome such a limitation, our method is based on an ap-proximated sequence alignment algorithm to search “sounds like ” sequences. Our approach does not depend on extra information from the ASR and outperforms up to 7 points the precision of state-of-the-art techniques in our experiments.