Output details
11 - Computer Science and Informatics
University of East Anglia
Inferring the structure of a tennis game using audio information
<22> We consider how audio events are related contextually for integration into audio-visual scene analysis. Most related work on content-based video retrieval or automatic analysis of tactics in games tends to be either audio-only or video-only, and uses only matches filmed at the same location: we use matches from multiple locations to test robustness of our techniques to different cameras, microphones, backgrounds etc.. This work is based on conference papers (10.1109/ICASSP.2010.5495935 and Proc. Interspeech 2010) which have over 15 citations on Google Scholar, and it was the inspiration for work by Geiger et al (Interspeech 2011) for modelling domestic noises.