Matlab based instructional tool for FFT parameter exploration

Although the Fast Fourier Transform (FFT) has been the staple of signal processing [Oppenheim(1998)] for many years, it is still frequently misapplied. In many cases, the confusion stems from misconceptions regarding the relations between time and frequency-domain parameters. Also, spectral leakage due to mismatches between the sample rate and the harmonic contents, and the choice of windowing technique, are a frequent cause of trouble. In this paper, we will first present a summary of how the various FFT parameters relate and can be chosen in a practical way, followed by a discussion on spectral leakage, windowing and zero-padding. Then, a MatLab-based tool is introduced to help visualize the relevant concepts. The tool allows the user to graphically evaluate the influence of the analysis parameters on harmonic signals, as well as on a custom dataset, such as a sound recording. This, then, allows the user to experiment with, and optimize, the FFT analysis parameters to enhance the resulting FFT spectrum, as well as visually compare the inverse of the spectrum produced with the original time-domain signal.