Acquire data in time domain (as a function of time)
Do FT to change to frequency domain
Since signal changes slowly - it has a low frequency
Noise changes quickly - has a high frequency
Keep only the low frequency components
Zero out high frequency
Inverse FT to get back original data
But no high frequency noise remains
Do variations of this on NMR all the time for noise reduction
or resolution enhancement
Built in to many programs (including excel?)