Deep Learning Based Active Noise Cancellation for Reducing UAV Propeller Sound

In this research, we investigate the application of Convolutional Neural Networks (CNN) utilizing deep learning methodologies to actively predict and cancel noise generated by UAV propellers. Our approach involves extracting features from spectrograms and waveform data to train a CNN model capable of predicting and generating counteractive noise signals for effective propeller noise cancellation. A comprehensive assessment of the proposed model was conducted, resulting in an accuracy of 94.5%, a precision of 93.3%, and a recall of 96.1%. Simulated testing in various scenarios showed promising results, demonstrating the model's capability in reducing drone propeller noise effectively. Furthermore, the model's performance re-mained robust under various simulated environmental conditions, indicating its potential for real-world applications. This research paves the way for more sophisticated noise reduction systems in UAVs, contributing to quieter and more environmentally friendly operations.