In our work, we have used two datasets that have been
acquired in context of sound detection for elder people.
Acoustic features have been identified to be potentially
useful for characterizing daily sounds and employed in our
work. Recently, ensemble methods have gained a lot of
attention due to their higher accuracy in classification [9].
The idea behind ensemble methodology is to vote the
individual predictions of classifiers and come up with a
model that outperforms the accuracy of every individual
classifier. This paper presents our investigations on popular
ensemble methods for sound classification. To our
knowledge, no such ensemble methods have been used for
such classification task. Ensemble methods have proved to be
effective in recognizing daily sounds as their results have
been compared with the individual classifier results. Our
study also makes a comparison of our proposed setup of
ensemble method results with previous results in literature
and shows a significant improvement over them.