Detection and classification of insects is important to avoid infestation in stored grains. Traditional methods of insect detections are visual detection, trapping and random sampling. These methods have their limitation in the terms of efficiency and can detect large sized insects but are unable to detect small insects and larvae. In this paper, the analysis of classification algorithms for detection of granaries insect is performed with the help of feature extraction using machine learning technique. Classification algorithms are applied on these extracted features with the help of MATLAB. The proposed method can distinguish hidden adult/larvae insects and their types. It differentiates insects sounds obtained from a grain silo based on the extracted features. The method of feature extraction and classification is robust, inexpensive, rapid and reliable. The results generated from datasets having 1500 instances in training dataset and 500/120 instances in testing dataset shows that SVM and KNN provide nearly same accuracy in the range of 84% to 90%, for decision tree the accuracy vary from 71% to 90%, for ensemble classifier the accuracy vary from 72% to 90%.