Biofeedback is currently considered as an effective method for medical rehabilitation. It aims to increase the awareness and recognition of the body's motion by feeding back the physiological information to the patients in real time. Our goal is to create an auditory biofeedback that aids understanding of the dynamic motion involving multiple muscular parts, with the ultimate aim of clinical rehabilitation use. In this paper, we report the development of a real-time sonification system using EMG, and we propose three sonification methods that represent the data in pitch, timbre, and the combination of polyphonic timbre and loudness. Our user evaluation test involves the task of timing and order identification and a questionnaire about the subjective comprehensibility and the preferences, leading to a discussion of the task performance and usability. The results show that the subjects can understand the order of the muscular activities at 63.7% accuracy on average. And the sonification method with polyphonic timbre and loudness provides an 85.2% accuracy score on average, showing its effectiveness. Regarding the preference of the sound design, we found that there is not a direct relationship between the task performance accuracy and the preference of sound in the proposed implementations.