Musical ensemble performance is a sophisticated form of joint action that showcases the remarkable human capacity for precise interpersonal coordination. Such coordination requires some degree of self-other merging while maintaining the distinction between self and other. Integration of information related to one's own and others' actions assists in representing shared goals and evaluating joint outcomes. At the same time, segregation between the effects of actions produced by self and others facilitates agency attribution and autonomous movement control. In this talk, I will present research addressing how self-other integration and segregation are balanced in the auditory domain during musical joint action. I will report findings from sensorimotor synchronization experiments employing controlled laboratory paradigms and naturalistic musical tasks, as well as related computational modelling work and neuroimaging studies. Results suggest that achieving optimal balance between self-other integration and segregation requires finely tuned internal models of self, other, and joint action outcomes.