In an embodiment, a transmitting device transmits a real-time media stream for a real-time communication session to a set of target devices. At least one of the set of target devices detects a media reception gap in the real-time communication session, and transmits a rewind request to the transmitting device to trigger the transmitting device to generate and provide a time-delayed version of the real-time media stream that includes media lost during the media reception gap. The transmitting device generates a rewind stream based on the rewind request and transmits the rewind stream to the at least one target device.