Extending Search System based on Interactive Visualization for Speech Corpora

Extending Search System based on Interactive Visualization for Speech Corpora Tomoko Ohsuga, Yuichi Ishimoto, Tomoko Kajiyama Shunsuke Kozawa, Kiyotaka Uchimoto, Shuichi Itahashi National Institute of Informatics, National Institute for Japanese Language and Linguistics, The Open University Gunosy Inc., National Institute of Information and Communications Technology, Tsukuba University 2-1-2 Hitotsubashi, Chiyoda-ku, Tokyo 101-8430, Japan 10-2 Midori-cho, Tachikawa City, Tokyo 190-8561, Japan Walton Hall, Milton Keynes MK7 6AA, United Kingdom Roppongi Hills Mori Tower, 6-10-1 Roppongi, Minato-ku, Tokyo 106-6125, Japan 4-2-1 Nukui-Kitamachi, Koganei, Tokyo 184-8795, Japan 1-1-1 Tennodai, Tsukuba, Ibaraki 305-8577, Japan osuga@nii.ac.jp, yishi@ninjal.ac.jp, tomoko.kajiyama@open.ac.uk shunsuke.kozawa@gmail.com, uchimoto@nict.go.jp, itabashi.shuichi.da@alumni.tsukuba.ac.jp Abstract This paper describes a search system that we have developed specifically for speech corpus retrieval. It is difficult for speech corpus users to compare and select suitable corpora from the large number of various language resources in the world. It would be more convenient for users if each data center used a common specification system for describing its corpora. With the “Concentric Ring View (CRV) System” we proposed, users can search for speech corpora interactively and visually by utilizing the attributes peculiar to speech corpora. We have already proposed a set of specification attributes and items as the first step towards standardization, and we have added these attributes and items to the large-scale metadata database “SHACHI”, then we connected SHACHI to the CRV system and implemented it as a combined speech corpus search system. Keywords: speech corpus, retrieval, visualization 1. Introduction Speech corpora are indispensable to speech research; several data centers of language resources have been set up worldwide to meet this demand by serving as a repository for language resources that include various speech corpora. They include the European Language Resources Association (ELRA), the Linguistic Data Consortium (LDC) in the U.S.A., the Chinese LDC / Chinese Corpus Consortium (CCC), the Speech Information Technology & Industry Promotion Center (SiTEC) in Korea, and the Speech Resources Consortium at the National Institute of Informatics (NII-SRC) / Language Resources Association (Gengo Shigen Kyokai; GSK) / Advanced Language Information Forum (ALAGIN) in Japan. The amount of data distributed from such data centers is very large. The diversity of the data gives users more freedom of choice, but it has become difficult to select suitable corpora for the intended purpose from the wide variety of corpora that have been made available. Although it is possible to search for these data from the website of each data center, the metadata of the corpus descriptions are not unified and also we cannot specify speech-specific conditions such as the recording environment in the search, so it is not easy for users to find the necessary corpus. Therefore, it would be more convenient for corpus users if the catalogue specifications of the corpora were standardized among all the various data centers and if the speech corpora could be retrieved by utilizing speech-specific conditions. We have already proposed an interactive visualization and search system for speech corpora called the “Concentric Ring View (CRV)” system, which simultaneously creates a visual display while performing data …

Content truncated to 500 words