"A good way to think about the rectangular window's `sinc` shape as it slides past the input signal is as a *probe* with a resolution defined by its mainlobe width. For example, in frame $k=12$, we see that the peak of the rectangular window coincides with the peak of the input frequency so we should expect a large value for $Z_{k=12}$ which is shown below. However, if the rectangular window were shorter, corresponding to a wider mainlobe width, then two nearby frequencies could be draped in the same mainlobe and would then be indistinguishable in the resulting DFT because the DFT for that value of $k$ is the inner-product (i.e. a complex number) of the two overlapping graphs.\n",
0 commit comments