python – Why FFT is different? Why so many frequencies?

Please help me here.

Here ‘hrir’ is a matrix which has 4 time responses. I am trying to build another matrix ‘c_f’ which will have FFT’s of these hrir’s. When I try to plot fft of one of the elements like- c_f[0,0], it gives me extra unnecessary frequency bins in the plot. Whereas when I simply take individual fft without matrix approach, and try to plot that (fft in code), it gives me a smooth plot. Of course the smooth plot is nicer and desirable. Why this happens? I mean, why unnecessary frequency data in the bad plot? Please see the python code and plots. If you see carefully, they have similar shape, just unwanted frequencies in undesired one. These Plots are nothing but magnitude spectrum and ‘plotfilter’ is a function which does that. I really want to continue with matrix approach..

code: