This research is part of a WAS-5 robot project to make it possible to assist music therapist in treating Autism Spectrum Disorder (ASD) subjects. One of the challenges faced by ASD subjects is social skills where tempo matching is an important factor when interacting with other people such as conversations and physical activities. This research aims to develop a system that trains the subject to match the music tempo played by WAS-5 with a feedback system. The system evaluates the tempo of the subject's movement and varies the tempo of the music in real-time to guide the subject in order to achieve tempo synchronization. Tempo has been successfully captured in this research through the use of Optical Flow with USB camera and achieved better accuracy using PID control. However, error is still present due to hardware limitations and human factors. Future works includes the fine tuning of the PID to lower error rate, hardware upgrade for better image acquisition, and preliminary experiments with a group of ASD subjects before implementation into actual therapy.