loss concealment

When you are listening to someone and a single syllable gets garbled, your brain often patches over it without you even noticing — you 'hear' what should have been there. Loss concealment is the receiver's version of that trick for real-time media. When a packet of audio or video goes missing and there is no time to ask for it again, the player does not just play silence or freeze; it generates a plausible substitute to fill the gap, so the listener or viewer ideally never perceives the hole.

How it papers over gaps depends on the medium. For voice, a simple method repeats the previous packet's waveform, fading it out so a short drop sounds like a tiny smear rather than a click of silence; better codecs synthesize a continuation of the speech sound. For video, a lost piece of a frame can be filled by copying the same region from the previous frame, since most of a picture changes little frame to frame. A complementary technique is forward error correction (FEC), where the sender proactively transmits a little redundant data so the receiver can reconstruct a lost packet from what did arrive — spending bandwidth up front to avoid relying on a too-late retransmission.

Why it matters: concealment is what lets VoIP and video calls survive the normal, modest packet loss of a best-effort network without becoming unbearable. It is the partner of the playout buffer: the buffer turns extreme lateness into loss, and concealment hides that loss. But be honest about its limits — concealment guesses, it does not recover the true data. At low loss it is nearly imperceptible; as loss climbs, the guesses become audible warbles or visible smears, and no amount of concealment can hide a connection that is dropping most of its packets.