Sound Research WIKINDX |
![]() |
Bakker, E. M., & Lewis, M. S. (2002). Semantic video retrieval using audio analysis. Lecture Notes in Computer Science, 2383, 271–277. Added by: Mark Grimshaw-Aagaard (4/20/05, 4:19 PM) |
Resource type: Journal Article BibTeX citation key: Bakker2002 Email resource to friend View all bibliographic details |
Categories: Typologies/Taxonomies Keywords: Semantic categorization, Silence Creators: Bakker, Lewis Collection: Lecture Notes in Computer Science |
Views: 35/1594
|
Abstract |
Semantic understanding of video is an important frontier in content based retrieval. In the research literature, significant attention has been given to the visual aspect of video, however, relatively little work directly uses audio content for video retrieval. Our paper gives an overview of our current research directions in semantic video retrieval using audio content. We discuss the effectiveness of classifying audio into semantic categories by combining both global and local audio features based in the frequency spectrum. Furthermore, we introduce two novel features called Frequency Spectrum Differentials (FSD), and Differential Swap Rate (DSR), that both model the shape of the spectrum. Added by: Mark Grimshaw-Aagaard |
Notes |
Searching video clips on the semantic properties of audio in those clips. If by semantic they intedn meaning, then their technique is not really semantic but rather one based on the means of sound production. Their categories include type of musical instrument, explosions speech etc.
Added by: Mark Grimshaw-Aagaard |