This project deals with methods for extracting features in signals, which then can be used in a machine learning approach for recognizing an aircraft in a signal. Noise is often unwanted and needed to be hold at a minimum, especially in areas close to where people are living and staying, for example around an airport. In that case, a maximum accepted sound pressure level will be set and the issue is now to automatically recognize when an aircraft sound is present in a recorded signal. A signal having an aircraft present will often have temporal variations, which we believe, when utilized in the analysis, improves the detection of an aircraft in a signal. We have investigated different feature extraction methods which utilizes temporal variation. We have used the Cyclic Analysis for obtaining information about (hidden) periodicities, in $\SI{1}{s}$ segments, which returns a Cyclic Spectral Coherence. We have also used a Harmonic Chirp Model (HCM) chirp pitch estimator for obtaining the instantaneous frequencies from an \textit{event} of an aircraft in a signal and used it in a flight parameter estimator for estimating flight parameters from a recording when an aircraft is passing by, by using parameters from its trajectory. The Cyclic Analysis obtained same classification error rates with a parametric classifier as the Mel-Frequency Cepstral Coefficients, used as baseline, did with a non-parametric classifier. We have shown the advantages of using the HCM chirp pitch estimator, we have tested the flight parameter estimator on synthetic signal and we have obtained great results of detecting aircraft within an event when an aircraft is passing by. All in all, we believe we have shown the benefits of using temporal variations in feature extraction from a signal when the purpose is to detect an aircraft within a signal.