Sound is considered as one of the distinguishing features for the classification of animals. Acoustic monitoring is one of the effective tools for ecological research and biodiversity assessment. Deep Learning has emerged as one of the effective techniques to deal with huge and complex datasets. It has a great ability to understand patterns in the data. It is used in various fields like image processing, robotics, etc. including bioacoustics. Despite advancement in the Deep Learning on bioacoustics classification, attaining high accuracy remains a significant challenge. In this paper, a combination of data augmentation and feature extraction techniques is proposed. The performance of DL models is analyzed in the context of data augmentation and feature extraction. Experimental results demonstrate significant improvements in classification accuracy compared to previous methods. The Deep Learning model achieves 93.71% accuracy on the ESC-50 datasets, representing significant improvements over previous methodologies.