When I am supposed to calculate an IR-, Raman-, or UV/Vis-spectrum, the calculations give me some x-y-values, e.g., Frequency-Intensity, or Energy-Oscillator Strength. Those values can either be printed as stick spectra or a sum of Gaussian-/Lorentz-distributions is being used to broaden the peaks to generate a more "natural" spectrum.
I've been asked a couple of times which broadening function to choose when, but actually never had a good answer/given explanation. The only thing that I wrote down quite some time ago, was that for IR- or Raman-spectra of solids a Gaussian broadening and for every other type of IR- or Raman-spectra a Lorentz broadening should be used.
Google searches for "gaussian or lorentz broadening" and "line broadening" did yield results that try to explain it, but I didn't find most of them really helpful.
So the question is: When do I choose the Gaussian or the Lorentz-functions for artificial broadening of IR-, Raman-, or UV-Vis stick spectra and why?