Abstarct: Information about the movement and type of an aircraft is usually collected by radar or electronic sensors, but this information may be unavailable due to lack of visual field. Detection and identification of the target by sound signals is very important in some cases because of geographic location, cost-effectiveness, and noise resistance. That is why another method is used to identify a target that is less disturbed by the conditions of the visual path. In this thesis the sound of an aircraft engine has been used to recognize the type of an aircraft. The audio signals received from the microphone are considered as inputs of this algorithm. In order to extract the characteristic of audio signal , the wavelet transform energy, Mel Frequency Cepstrum Coefficients (MFCC), Real Cepstrum Coefficients, linear predictive coefficients (LPCs) and spectral line frequency (LSF) have been used. Then classification of the sound signal of the aircraft is carried out for four classes of an audio data by artificial neural network classifier. To evaluate the performance of the detection system, we added noise to the signal with different signal-to-noise ratio, and compare the results with each other. Finally, we introduce a method for improving the system performance, which is used wavelet transform to eliminate the noise, and combination of LSF and real cepstrum coefficients are used as a feature extraction method. The data classification rate was %93.11 and the error rate was 0.0451.