Drones have become affordable to common people in this decade, it raises the number of private drone users. There are lot of advantages of private usage of drones, at the same time it can create problems as well, from challenging the privacy of individuals to threatening the national security.  Thus there is a need for drone detection systems not only in high security area but also in private premises to alert any potential threats. There are numerous researches going on in drone detection that investigate several detection methods, such as RADAR based detection, visual detection, RF sensing and acoustic detection. Acoustic detection is one of the popular method since it does not require line of sight, can work in low visibility area and it is low cost. However, most of the acoustic based detections adapt methods from speech based literature. Effective implementation of a drone detection system requires careful study on the sound which drone produces. Sound of drones may vary with type of motors, number of motors, shape and number of propellers and moving orientation. So for our study, two sets of drone sounds were collected. One set was collected at the recording theater with one drone, DJI Mavic mini in all possible orientations and operations and another set was collected from internet which include the sounds of commonly used drone models. Sounds collected in recording theatre were used to study the features of the sound. There are other sounds possible at the detection field and they may sound similar to the drone such as vehicles, crowd, lawn mowers, airplane and blowing air. It is important to find decisive features to separate those sounds from drone sound. For that study, sounds that are possible at the detection field also were collected. Comparison studies were conducted in both time domain and frequency domain to find discriminative features. SVM based detection model will be proposed to detect drones using acoustic signal.