Most of the conventional electric prostheses mainly focus on the grip function and there are few prostheses aiming to play musical instruments. This study aims to develop a myoelectric prosthetic hand control system to play recorders and a training system suitable for the recorder performance. Each finger driven by a shape memory alloy (SMA) actuator is controlled based on the forearm gesture recognition such as grasping and flexion using a wireless EMG sensor and also a visual feedback training system for recorder performance is developed to improve the proficiency level of the EMG operation and recorder performance. Some experiments are conducted to show the effectiveness of the proposed systems.