What are the different kinds of FFT Processing Possible?

Some things I have tried except a few ones:

- keep only magnitudes higher than a threshold (spectral gating), lower than a threshold (only noise remains in sounds with acoustic origin), or between two thresholds.
- split with threshold(s) and send to differents destinations (to different loudspeakers = spatialization). You can also send to different circuits of transformation ;
- keep only a given number of the highest magnitudes.
- reduce/increase contrast of magnitudes.
- reverse spectrum (a bit deceiving as FFT works in a linear scale).
- increase or reduce the magnitudes variations between consecutive frames.
- remove magnitudes below/over a specific band number (high-slope highpass/lowpass filtering) or between/outside two band numbers (notch/bandpass).
- remove/keep one in two/three/four/n bands (kind of comb filtering).
- multiply each magnitude by a value stored in a buffer~ (graphic filter)
- split according to band number (crossover filter, spatialization)
- apply an offset on bands (frequency shift)
- keep only n bands and copy them as many times as needed to fill the whole spectrum between O Hz and Nyquist
- switch bands (kind of spectral blurring), for instance odd and even or randomly
- spectral interlacing (take m bands from one sounds, n next bands from another…)
- freeze
- substract one sound’s magnitudes from those of another (kind of inverted convolution…)
- shift phases (to increase spatial width for instance).
- detect transients
- detect pitch
…