Existing media playout scheduling (MPS) schemes usually focus on selecting and scheduling packets according to optimized Internet media metrics, which are only partially relevant to the subjective human perception in the interactive system. The MPS design challenges are two-fold. First, human preferences are concurrently dominated by multiple quality attributes of the streaming media whose perceptual tradeoffs were not well understood, so they were not used as an integral part of an efficient MPS design. Second, people's perceptions can be impacted by the flicker effect caused by Internet dynamics and the resulting MPS adaptations. In this paper, we propose a new and adaptive perception-based MPS scheme to deliver high-quality real-time interactive multimedia. We first investigate the perceptual tradeoffs among the multi-modal bundle streaming qualities in a real Internet environment. We then present ourMPS design that finds the bundle quality tradeoffs, while minimizing flicker degradations. Evaluation results show the performance of our MPS scheme.