In the last paragraph of this article, the author gives an example of the DFT of a song and his window's resonant frequency. As he points out (and it is a bit intuitive), if he listens to this song, the window is not likely to break, despite the song's spectrum being centered around the natural frequency of the window.
The question is, if we are given a range of resonant frequencies (the "dangerous" range around the resonant frequency), can we decide whether "the window would break", so to speak, based on the DFT alone, without a physics model (and therefore regardless of the problem, i.e. a window, a vibrating wire, etc.)?
Also, if we have two different signals and we compute their DFTs, how does the real-world physical effect of those signals (say, on a window) compare to their difference in amplitude for the resonant frequency? That is, if the amplitudes are equal, are the effects on the window's vibrations also equal? If one amplitude is half the other one, are the vibrations proportional the amplitudes' ratio (1/2)?