The ultimate goal of this study is to develop a robot that can play an alto recorder so that it sounds like a human is playing it. We have used unsteady flow rate measurements and control techniques in our research. It is particularly difficult for a robot to perform special musical effects, such as vibrato, tremolo, and tonguing. We present a method that models eight elements of the flow rate data in order to create a versatile vibrato. An evaluation experiment was conducted, and the proposed versatile vibrato was confirmed to be closer than monotone vibrato to human-performed vibrato.