SIGNAL LAB · MEASURE · COMPARE
Jitter buffer: arrivals versus playout deadlines
Trace fixed-rate packets through variable network delays and see exactly which ones miss their playout slots.
Exact evidence inspector
Read the selected state numerically. This includes the details behind the chart rather than a preset verdict.
Show computed state
The sender clock must stay regular
Sixty packets are emitted exactly twenty milliseconds apart. Each packet has an identifier, send timestamp, deterministic network delay and fixed playout deadline. Network delay is forty milliseconds plus a seeded nonnegative variation. Extra buffering postpones every playout deadline by the same amount; it does not change the sender rate or repeatedly wait for a packet to become old enough. A packet arriving on or before its deadline can play. A packet arriving afterward is discarded for that slot, and a lost packet cannot be recovered by waiting longer.
Change one source of delay
Extra playout buffer adds between zero and two hundred milliseconds to the base forty-millisecond deadline offset. Maximum added network delay controls the upper bound of the seeded delay variation; it is not the RTP interarrival-jitter statistic. Lose every Nth packet removes each matching packet from the arrival sequence; zero disables loss and one loses every packet. Every setting resets the same sixty-packet fixture. The diagram shows ten packet rows around the current sender position: solid squares are sends, circles are arrivals and outlined markers are deadlines. The red cursor is model time. The exact inspector includes all sixty packets, not only the visible rows.
Trade latency for fewer misses
Set added network delay to zero, loss to zero and buffering to zero. Every packet arrives exactly on its deadline and all sixty play; arrival wins the tie. Raise added delay to eighty while leaving buffering at zero. Delayed packets miss their fixed slots even if they eventually arrive. Now set buffering to eighty: all packets should arrive in time in this bounded fixture, at the cost of eighty additional milliseconds before playout. Set loss to five and run again; twelve packets still miss because buffering cannot manufacture missing payloads. Compare these results using the same seed rather than a new random traffic sequence.
Inspect the timeline and accounting
Play maps the entire packet experiment to sixty inspection seconds; Pause freezes it, Step adds one inspection second, Restart rewinds, Speed changes viewing pace and Inspect time scrubs the record. Restore defaults resets all parameters. Played plus missed counts only deadlines already reached, so the denominator is neither packets sent nor packets received. Late discards count delayed arrivals observed so far; a missed packet may not have arrived yet. Buffered packets have arrived but their slots are still in the future. This fixed-delay teaching receiver omits adaptive jitter estimation, clock drift, packet-loss concealment, retransmission, decoding and actual RTP sequence handling. No network traffic or sound is produced.