Yan Zhang, Dan jv Lv, Ying Lin
Environmental audio classification has been the focus in the field of speech recognition. The paper proposes using ensemble learning to perform environmental audio classification. Ensemble learning is to train multiple component learners and combine their predictions to enhance the accuracy of classification. The paper explains the ensemble algorithms and strategies, and the comparison and analysis of classification results are given which obtain by employing the single classification algorithm, Bagging, Boosting, Random Forest and MCS ensemble classifiers for the given environmental audio data. The experimental results show that the ensemble technology can effectively improve the performance of environmental audio data classification. Even under the fewer number of the training examples, it provides a kind of effective approach to guarantee the performance and generalization of classification.