In time axis normalization I have used forward/inverse Fourier
transform on handwriting (pen-tip kinematics) signals with success. Since
computers are much faster than they used to be, it is not absolutely necessary
to perform an FFT with its power of 2 restriction on sample size and the
consequent biases introduced by the adding of zeroes. Using Slow Direct Fourier
Transform, one can select "any" spacing in the time and frequency domain.
Sample records were made circular by linear detrending.

Lambert Schomaker.