Realistic peak width and peaks convolution

Hello, I'm a new pyPENELOPE user. I'm trying to simulate some spectra of some real sample that I've already analized with EDS. From the results (I've attached a image in the end of this post for a better understanding, it's a simple specimen of 100% gold) it's possible to see that peak heights ratios are almost the same between the real spectrum (up) and the simulated one (down), but the simulated one is not realistic because the signals are lines and not bands, and all different signals are not convoluted like a real spectrum.
I'm wondering if it would be possible simulate spectra with more realistic appearence maybe adding some information of the detector used.

Thank you,

Walter

https://s22.postimg.io/785i94eup/image.jpg