Microbubbles are expected as carriers of acoustic targeted drug/gene delivery therapy. To realize the therapeutic system, we have proposed a technique that controls microbubble behavior in blood vessels using ultrasound emitted from the body surface. To apply this technique in vivo, robotic positioning of therapeutic devices on body surface is required. For this purpose, though we have developed a robotic system using parallel link robot, the motion of the diagnostic probe positioning was not included in the previous system. To carry out the cooperation between ultrasound diagnosis and therapy at the same time, we have developed a gripping mechanism of multiple sound sources. In this paper, we introduce the prototype of this mechanism and perform evaluation experiment of positioning accuracy. The results demonstrated that the system has a great potential to realize accurate positioning with 1.5mm in translation and 2.5deg in rotation.