This package implement the random regression forest algorithm for audio event detection in continuous streams. This algorithm was used in our following works:
[1]. Huy Phan, Marco Maass, Radoslaw Mazur, and Alfred Mertins, Acoustic Event Detection and Localization with Regression Forests, Proc. 15th Annual Conference of the International Speech Communication Association (INTERSPEECH 2014), Singapore, pp. 2524-2528, September 2014
[2]. Huy Phan, Marco Maaß, Radoslaw Mazur, and Alfred Mertins, Random Regression Forests for Acoustic Event Detection and Classification, IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLP), vol. 23, no. 1, pp. 20-31, January 2015
[3]. Huy Phan, Marco Maass, Radoslaw Mazur, and Alfred Mertins, Early Event Detection in Audio Streams, Proc. IEEE International Conference on Multimedia and Expo (ICME 2015), Turin, Italy, pp. 1-6, July 2015
[4]. Huy Phan, Marco Maass, Lars Hertel, Radoslaw Mazur, Ian McLoughlin, and Alfred Mertins, Learning Compact Structural Representations for Audio Events Using Regressor Banks, Proc. 41st IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP 2016), Shanghai, China, pp. 211-215, March 2016
Please note that the implementation is not optimized anyway. In addition, source code for the feature set used in the paper can be found here: https://github.com/pquochuy/Audio-Event-Features
The script main_forest.m gives a brief tutorial how to use the package for training and testing. If you got problems or questions regarding to this package, please email me at phan{at}isp.uni-luebeck.de
If you use this package for your work, please cite the following paper:
Huy Phan, Marco Maaß, Radoslaw Mazur, and Alfred Mertins, Random Regression Forests for Acoustic Event Detection and Classification, IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLP), vol. 23, no. 1, pp. 20-31, January 2015