Study and application of acoustic information for the detection of harmful content, and fusion with visual information
This thesis aims at investigating and developing techniques for content-based segmentation and classification of multimedia files, based on audio information. Emphasis has been given to analyzing the content of films based on audio information. In addition, part of the thesis is focused on the detection of audio classes related to violent content (e.g., gunshots, screams, etc).
File Type:
pdf
File Size:
3 MB
Author:
Giannakopoulos, Theodoros
Supervisors:
Sergios Theodoridis
Institution:
University of Athens
Keywords:
Audio analysis, violence detection, multimedia analysis, speech - music discrimination, audio classification, bayesian networks