 |
2003 ISCA Workshop on
Multilingual Spoken Document Retrieval
(MSDR2003)
Hong Kong
April 4-5, 2003 |
 |
Speaker Naming System by Associating Speech and Speaker Recognition Results
Y. Ariki (1), M. Nishida (2)
(1) Ryukoku University, Seta, Otsu-shi, Shiga, Japan
(2) PRESTO, Japan Science and Technology Corporation,
Sakyo-ku, Kyoto, Japan
In this paper, we propose a system which can associate
person names to individual speaker section. For this purpose,
the automatic speaker segmentation is carried out
utilizing online speaker modeling and speaker verification
techniques. Key phrases and person names are also
extracted by speech recognition. After this speaker segmentation
and speech recognition, the person name is
associated to the corresponding speaker sections based
on simple rules. This method can be expanded to video
clip retrieval by a query specifying both of a person name
and topic contents.
Full Paper
Bibliographic reference.
Ariki, Y. / Nishida, M. (2003):
"Speaker naming system by associating speech and speaker recognition results",
In MSDR-2003, 61-66.