This paper proposes the CPG synergy model--a biomimetic rhythm generator model based on central pattern generators (CPGs) and muscle synergy theory to enable evaluation of rhythmic motions with non-stationary characteristics such as human finger tapping movements. The model consists of multiple CPGs to approximate the complex rhythmic movement of humans, and has the potential to allow evaluation of abnormal movements in patients with motor function impairments such as Parkinson's disease (PD). To verify the validity of the proposed model, comparison experiments were conducted using model parameters (i.e., synergies, weight coefficients and time-shift parameters) extracted from finger tapping movements performed by individuals in a healthy subject group and a PD patient group. The results showed that the number of synergies, the second moment of synergy shapes and the coefficient of variation of maximum weight coefficients show significant differences for each subject group, and indicated that the model could be used to evaluate irregular rhythmic movements as well as regular ones.