Digital Signal Processing Reference
In-Depth Information
Many occurrences of filler words such as “eto”, which roughly
corresponds to “ummm” in English, are attested in the corpus; humans
typically do not inject these filler words into their dialogs with an
HMI.
To address these problems, we are planning to use the first prototype
dialog system (based on the human-human spoken dialog corpus) to collect
human-machine dialogs, which will be used to improve the dialog system in
subsequent incremental refinement.
ACKNOWLEDGEMENTS
The authors wish to thank Mr. J. Watanabe of Hitachi Ltd for his valuable
comments. The authors are also grateful to Dr. M. Araki, Assistant Professor
of Kyoto Institute of Technology, for his ongoing support of the spoken
dialog corpus development.
REFERENCES
[1]
E. Nyberg, T. Mitamura, P. Placeway, and M. Duggan and N.Hataoka, “DialogXML:
Extending VoiceXML for dynamic dialog management,” Proc. of Human Language
Technology Conference, 2002
C. Doran, J. Aberdeen, L. Damianos, and L. Hirshman, “Comparing Several Aspects of
Human-Computer and Human-Human Dialogues”,
[2]
Proc. of Second SIGdial Workshop
on Discourse and Dialog, Aalborg, Denmark.
C. Cieri, “Resources for Robust Analyses of Natural Language”, Conference ROMAND
2000, Lausanne, France, 2000
NARA
[3]
[4]
INSTIUTE of SCIENCE and TECHNOLOGY,
“Morphological
Analyzer
ChaSen”, http://chasen.aist-nara.ac.jp/
Y. Obuchi, E. Nyberg, T. Mitamura, M. Duggan, S. Judy and N. Hataoka, “ Robust Dialog
Management Architecture Using VoiceXML for Car Telematics Systems,” Proc. of
Workshop on DSP in Vehicular and Mobile Systems, 2003,“ Proc. of Workshop on DSP
in Vehicular and Mobile Systems, Nagoya, Japan, 2003.
M. Tateishi, I. Akahori, S. Judy, Y. Obuchi, T. Mitamura, and E. Nyberg, ”A Spoken
Dialog Corpus for Car Telematics Services,” Proc. of Workshop on DSP in Vehicular and
Mobile Systems, Nagoya, Japan, 2003.
W3C, “Voice Extensible Markup Language (VoiceXML) Version 2.0 Working Draft,”
http://www.w3c.org/TR/voicexml20/
[5]
[6]
[7]
