Latency is the delay time for data to travel from sender to receiver. In telephone communications, it is experienced as the time gap between speaking and the other party hearing your voice, making it one of the metrics affecting call quality. ITU-T Recommendation G.114 states that if one-way delay is kept below 150 ms, most applications experience essentially transparent interactivity.
The perception of latency degrades in stages. At 0-150 ms, natural conversation flows and delay is barely noticeable. The larger the delay grows beyond 150 ms, the more noticeable delayed responses and talk-over (both parties starting to speak at once) can become. G.114 treats one-way delays above 400 ms as unacceptable for general network planning. Calls routed via a geostationary satellite (about 36,000 km altitude) take about 260 ms one way for propagation between earth stations alone (a G.114 planning value), creating a distinctive conversational "gap."
VoIP call latency is the sum of multiple factors. Codec encoding/decoding (under 1 ms for G.711 and 25-35 ms for G.729 in the G.114 appendix tables), propagation across the network (about 5 ms per 1,000 km of optical fiber), and waiting time in the receiving jitter buffer (which absorbs variations in packet arrival times) all add up. The G.114 appendix expects mouth-to-ear delays below 150 ms for VoIP calls on many routes within a region of 5,000 km or less, and estimates about 225 ms for inter-regional calls of around 10,000 km.
Jitter is a related concept that should be kept distinct from latency. While latency is the "absolute value of delay," jitter refers to the "variation in delay." If latency is constant, call quality remains stable, but high jitter causes irregular packet arrival intervals, resulting in audio dropouts and distortion. QoS settings must address both latency minimization and jitter suppression. Even with sufficient bandwidth, router processing delays and network congestion can increase latency, making overall network design critical for VoIP deployments.