Linefeed insertion into Japanese spoken monologue for…换行插入日语口语独白.pdfVIP

Linefeed Insertion into Japanese Spoken Monologue for Captioning Tomohiro Ohno Masaki Murata Shigeki Matsubara Graduate School of Graduate School of Information Technology Center, International Development, Information Science, Nagoya University, Japan Nagoya University, Japan Nagoya University, Japan matubara@nagoya-u.jp ohno@nagoya-u.jp murata@el.itc.nagoya-u.ac.jp Abstract linefeed insertion for the displayed text is desired so that the text becomes easy to read. To support the real-time understanding of Until now, there existed few researches about spoken monologue such as lectures and how to display text on a screen in automatic cap- commentaries, the development of a cap- tioning. As the research about linefeed insertion, tioning system is required. In monologues, Monma et al. proposed a method based on pat- since a sentence tends to be long, each terns of a sequence of morphemes (Monma et sentence is often displayed in multi lines al., 2003). However, the target of the research is on one screen, it is necessary to insert closed-captions of Japanese TV shows, in which linefeeds into a text so that the text be- less than or equal to 2 lines text is displayed on comes easy to read. This paper proposes a screen and the text all switches to other text at a technique for inserting linefeeds into a a time. In the work, the highest priority concept Japanese spoken monolo