Artificial Intelligence, data-driven decision techniques, and the use of drones as swarm technology have rendered drones a potential weapon for mass destruction. Their small size and ability of autonomous flight let them evade Radar, Radio Frequency, and Vision-based detection techniques, hence, increasing their threat to military sensitive areas. However, Acoustic Based Drone Detection has been proven to remain effective as it utilizes drones’ acoustic signature to detect it. This paper proposes a low-cost and highly reliable Acoustic-Based Drone Detection System that employs machine learning at its core. The drone database was acquired from GitHub, and it contained acoustic signatures of 7 different drones. The environmental signals were collected from the BBC sound database and YouTube. 26 Mel Frequency Cepstral Coefficients (MFCCs) were extracted from the audio signals and the data was labeled as ‘Drone’ and ‘No-Drone’. Before training, clustering was performed to provide ground truth that the drone database contains acoustic signatures from 7 different drones. MFCCs were fed into Balanced Random Forest (RF) and MLP algorithms. The preliminary results of both algorithms were comparable with a combined average F-score of 0.92 (on training data) and 0.79 (test data) which indicates that our proposed system can reliably detect drones from their acoustic signature in real-time environments.