Continuous media applications have an implied temporal dimension, i.e. they are presented at a particular rate for a particular length of time and if the required rate of presentation is not met the integrity of these media is destroyed. We present a set of language constructs suitable for the definition of the required QoS and the real-time dimension of the media that participate in a multimedia application. We propose a new modeling scheme where we can represent every different presentation by a periodic task and then we describe a method that provides low-level support to these language constructs. The emphasis of the proposed strategy is given on deterministic guarantees and can be considered as a next step for the design and the implementation of predictable continuous media applications over a network.