In fact the this code is supposed to be a phenomenological model of surface electromyograms. You are not able to test your sEMG decomposition algorithm because there are actually no individual MUAPs simulated with this code (only "bursts" corresponding to the MUAP summation).
I have two alternatives for you:
1) Use one of the several synthetic signals from the SENIAM project (http://www.seniam.org) that comes with the CD (unfortunately not for free); or
2) Implement some simple model of MUAP trains for simulation. I recommend the following model for easy implementation: Rosa et al. A novel electromyographic signal simulator for muscle contraction studies. Computer Methods and Programs in Biomedicine 2008; 89: 269-274.
Let me know if I can be of any help.