Towards robust spoken dialog systems using large-scale in-car speech corpus

Yamaguchi, Yukiko; Hayashi, Keita; Ono, Takahiro; Kato, Shingo; Irie, Yuki; Ohno, Tomohiro; Murao, Hiroya; Matsubara, Shigeki; Kawaguchi, Nobuo; Takeda, Kazuya

Towards robust spoken dialog systems using large-scale in-car speech corpus

Yukiko Yamaguchi, Keita Hayashi, Takahiro Ono, Shingo Kato, Yuki Irie, Tomohiro Ohno, Hiroya Murao, Shigeki Matsubara, Nobuo Kawaguchi, Kazuya Takeda

We have been studying various topics by using a large-scale corpus, which was built at CIAIR, to construct a robust and practical spoken dialogue system. The CIAIR project has developed a data collection vehicle and collected about 179 hours of multi-modal data in total. We have transcribed the speech data by about 800 subjects, and annotated speech intentions, dependency structures, dialogue structures to the text data. We are continuing various research using the annotated data, such as speech. Intention understanding and speakers knowledge acquisition. In this paper, we introduce our research activities, and present the various fruits of the in-car speech corpus.