Analysis of acoustic data with intelligent computational methods

Nowadays, as the use of aircraft is rapidly increasing, the importance of acoustic analysis increases with the detection of these vehicles. Aircraft such as helicopters and drones emit sounds in a certain frequency range, and the analysis of these sounds allows the flight characteristics of the vehicles to be understood. Both types of vehicles produce different sound profiles due to their different aerodynamic structures and engine systems. This thesis aims to analyze drones and helicopters with acoustic data. The study classifies the characteristic sound features of each vehicle type using acoustic parameters such as Mel-Frequency Cepstral Coefficients (MFCC) and Zero-Crossing Rate (ZCR) obtained from sound files. A classification model that can distinguish between drone, helicopter and environmental sounds was developed with the data obtained from a total of 546 sound files. The training process was evaluated with performance criteria based on the complexity matrix. The acoustic analysis performed using MATLAB software revealed the acoustic signatures of both vehicle types, and a model that minimizes the effect of environmental noise was proposed with this software. The results show that acoustic analysis can be an important tool in applications such as identification and tracking of drones and helicopters. In addition, the model, which minimizes the effect of environmental noise and detects only target sounds, shows the potential to be used in security and monitoring technologies. This study provides an important step in the classification of aircraft and environmental sounds with acoustic data.