Recherche par le contenu dans des documents audiovisuels multilingues

ABSTRACT. We present in this paper an approach based on the use of the International PhoneticAlphabet (IPA) for content-based indexing and retrieval of multilingual audiovisual documents.The approach works even if the languages of the document are unknown. It has been validatedin the context of the “Star Challenge” search engine competition organized by the A*STARAgency of Singapore. Our approach includes the building of an IPA-based multilingual acousticmodel and a dynamic programming based method for searching document segments by “IPAstring spotting”. Dynamic programming allows for retrieving the query string in the documentstring even with a significant transcription error rate at the phone level. The methods that wedeveloped ranked us as first and third on the monolingual (English) search task, as fifth on themultilingual search task and as first on the multimodal (audio and image) search task. MOTS-CLES : Recherche audio, Multilingue, Alphabet Phonetique International, ProgrammationDynamique, Star Challenge