The strategy the central nervous system utilizes to produce movements in the face of multiple degrees of freedom available has been a subject of study for the past few years. Of the possible mechanisms, the muscle synergies - stereotypical coordinated patterns of muscle activity elicited by dedicated networks have been suggested to be the building blocks. Based on this hypothesis, several algorithms have been proposed to discern these synergies from the recorded electromyographic signals (EMG). In the proposed model, the synergies are treated as filters (IRFs) that take as input any arbitrary non-negative signal. That is, the EMG is seen as a convolution mixture of synergies and corresponding inputs.