Audio events in soccer videos classification using deep belief network

We propose an approach for the classification of audio concepts in sport videos using deep belief networks (DBNs), a probabilistic neural network with several hidden layers. Comparison with support vector machine (SVM) classifiers has been carried on, showing that our preliminary results are promisingly comparable to the state-of-the-art.

Dataset description
1284 .wav files: 1ch 16k 2s clips
[filename].label files: label for [filename].wav clip

label values:

• 0: SILENCE
• 1: SPEECH_ONLY
• 2: SPEECH_OVER_CROWD
• 3: CROWD_ONLY
• 4: EXCITED

Please, if you use the dataset cite our papers as follows:

@inproceedings {icme2009,
       Acceptance = {Oral Acceptance Rate 22\%},
       Address = {New York, NY, USA},
       Author = {Ballan,Lamberto and Bazzica,Alessio and Bertini,
                 Marco and Del Bimbo, Alberto and Serra,Giuseppe},
       Booktitle = {Proc. of {IEEE} International Conference on Multimedia & Expo (ICME)},
       Date-Added = {2009-07-09 09:28:17 +0200},
       Date-Modified = {2010-09-13 11:21:06 +0200},
       Doi = {http://dx.doi.org/10.1109/ICME.2009.5202537},
       Keywords = {Audio analysis, Deep Belief Networks},
       Month = {July},
       Pages = {474--477},
       Title = {Deep Networks for Audio Event Classification in Soccer Videos},
       Url = {http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5202537},
       Year = {2009},
       Bdsk-Url-1 = {http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=5202537},
       Bdsk-Url-2 = {http://dx.doi.org/10.1109/ICME.2009.5202537}
}