Performance

Jitter Buffer

Enter jitter, packetization and latencies and I compute the recommended jitter-buffer depth, packets queued and total mouth-to-ear latency, rated with the G.114 threshold (≤150 ms good).

redzilla.cl — jitter

Buffer depth

—

Packets queued

—

Mouth-to-ear latency

—

Buffer —

Packets —

Mouth-to-ear —

ITU-T G.114 rating

— —

—

Mouth-to-ear delay breakdown

Component	ms	%

How it is computed · buffer, packets and G.114

1. Buffer depth: k × jitter, with a floor of 2 packets (2 × ptime). With k = 2 the buffer absorbs twice the observed jitter.

2. Packets queued: ceil(depth / ptime). The buffer fills with whole frames, so it rounds up.

3. The buffer adds its depth as fixed delay. Mouth-to-ear latency = network one-way + buffer + codec + ptime.

4. ITU-T G.114 (one-way delay): ≤ 150 ms good, 150–400 ms acceptable, > 400 ms poor for most voice applications.

Runs locally in your browser · no sign-up · nothing leaves your browser

How it works

The calculator sizes the jitter buffer for a VoIP call or video stream and estimates the resulting mouth-to-ear latency. You enter the network jitter, the packetization interval (ptime, typically 20 ms for voice), the one-way network latency and the codec delay; from those it derives the recommended buffer depth, how many packets it queues and the total delay the listener perceives.

The criterion is the usual one in voice engineering: buffer depth = k × jitter with a floor of 2 packets (2 × ptime), queued packets = ceil(depth ÷ ptime) and mouth-to-ear = one-way network + buffer + codec + ptime. The total is evaluated against the ITU-T G.114 recommendation for one-way delay: up to 150 ms is good, 150-400 ms acceptable and above 400 ms poor for conversation.

Example: LAN VoIP with G.711 and 20 ms of jitter

With 20 ms jitter, 20 ms ptime and factor k = 2: depth = max(2 × 20, 2 × 20) = 40 ms.
Queued packets: ceil(40 ÷ 20) = 2 packets.
Mouth-to-ear: 50 (network) + 40 (buffer) + 15 (codec) + 20 (ptime) = 125 ms.
G.114 verdict: good, because 125 ms stays under the 150 ms threshold.

Frequently asked questions

How many milliseconds of jitter buffer do I need for VoIP?

The rule of thumb is 1.5 to 2 times the measured jitter, with a minimum of 2 packets (40 ms at a 20 ms ptime). With 30 ms of jitter, a 60 ms buffer absorbs the variation without adding unnecessary delay. An oversized buffer does not drop packets, but it adds latency directly to the conversation.

What happens if the jitter buffer is too small or too large?

If it is too small, packets that arrive late are discarded and the audio sounds choppy (equivalent to packet loss). If it is too large, every millisecond of buffer adds to the mouth-to-ear delay and speakers start talking over each other. The sweet spot covers the real jitter with the least possible delay.

What is the maximum acceptable latency for a call according to G.114?

ITU-T G.114 recommends keeping one-way delay at 150 ms or less for good conversational quality; between 150 and 400 ms it is acceptable but the latency is noticeable, and above 400 ms conversation becomes difficult, with overlaps and double talk. A geostationary satellite link (about 300 ms one-way) starts out in the high-acceptable zone.

What is the difference between latency and jitter?

Latency is how long a packet takes to arrive; jitter is how much that time varies between consecutive packets. A network can have high but stable latency (low jitter) and sound fine with a small buffer, while a network with high jitter needs more buffering even if its average latency is low.

Was this tool useful?

Disclaimer

We take great care to keep every tool accurate and review it thoroughly; even so, we can't guarantee it is free of errors or take responsibility for how the results are used. We recommend double-checking anything critical.

Found an error? Let us know →