This paper describes a variant of the actor model suited to the development of multimedia systems. The actor model centers on non-over killing concurrency and customizable constraint-directed scheduling. A distributed multimedia system consists of a collection of actors interacting to one another to fulfil a common goal, e.g., interactive videoconference, video recording on-demand, etc. Timing synchronization rests on reflective actors (QoSsynchronizers) which filter actor messages and schedule them according to application-dependent QoS parameters. Requirements of a multimedia system can be formalized by Time Stream Petri Nets. In order to support QoS constraints analysis and validation, media actors are then prototyped under simulation by exploiting a flexibility of the adopted actor model to operate transparently under real or virtual time. The transition from the modeling, to analysis, down through to the implementation stages is therefore seamless. The paper describes the application of the methodology to the development of a lip-sync filter for live or on-demand multimedia sessions and reports some collected experimental results.