It is a common practice to apply windowing function, such as Hann or Hamming, to a time domain signal before FFT, in order to reduce spectral leakage. Often, we do 1) Windowing, 2) FFT, 3) frequency domain processing, such as filtering, then 4) Inverse FFT. My questions are: before inverse FFT, do we need to apply a windowing function in frequency domain as well? If we do, how?
Thanks in advance.