Electromyographic (EMG) signals exhibit complex interference patterns that comprise several single motor unit action potentials (SMUAPs). Evidence of a model that can generate EMG signals and considers intrinsic characteristics, such as long-range dependence (LRD) or shortrange dependence (SRD), or that supports the study of pathology-related signals is lacking. Therefore, the present study aimed to develop an EMG signal generator based on SRD or LRD derived from firing patterns. We used a dynamic model to parameterize up to 15 SMUAP waveforms of real EMG signals extracted from a database. Then, we used relative appearance rates for some signals based on the number of SMUAPs to generate the latter randomly. Furthermore, we complemented our model by generating a random firing pattern. The synthetic reconstruction of the signals indicated a displacement compared with their respective firing patterns, with the highest error rate being 4.1%. The model of the EMG signal generator in its current state could be useful for a specialist who intends to study the behavior of the signals, starting with the exploration of synthetic signals and then proceeding to the real signals.