Enhanced video handling based on audio analysis

Author

Minami, Kenichi ; Akutsu, Akihito ; Hamada, Hiroslni ; Tonomura, Yoshinobu

Author_Institution

NTT Human Interface Labs., Kanagawa, Japan

fYear

1997

fDate

3-6 Jun 1997

Firstpage

219

Lastpage

226

Abstract

Soundtracks of videos contain a rich source of content-based information. In this paper, we propose an audio-based approach to video indexing and handling. Audio data is analysed by means of frequency analysis, and music and voice are independently detected even if they occur together. The method is implemented on a system called Video in Time as an example of creating reasonable condensed versions of dramas or movies by excerpting meaningful video segments. Users can select the desired replaying time from several different levels, depending on how much time can be afforded for viewing. Detection rates for music and voice are evaluated and experiences with the system are mentioned

Keywords

audio-visual systems; data analysis; image segmentation; indexing; multimedia computing; music; speech processing; Video in Time; audio analysis; content-based information; detection rates; drama; enhanced video handling; frequency analysis; movies; music; time; video indexing; video segments; video soundtracks; voice; Data mining; Feature extraction; Humans; Indexing; Laboratories; Layout; Motion pictures; Speech analysis; Telegraphy; Telephony;

fLanguage

English

Publisher

ieee

Conference_Titel

Multimedia Computing and Systems '97. Proceedings., IEEE International Conference on

Conference_Location

Ottawa, Ont.

Print_ISBN

0-8186-5530-5

Type

conf

DOI

10.1109/MMCS.1997.609596

Filename

609596