Low average latency alone does not guarantee that voice calls will be uninterrupted. Changes in packet latency over time also affect the experience.

Basic approach

Jitter describes delay variability in everyday use; tools can calculate this with different formulas. Before comparing the jitter number in two instruments, it is necessary to know the sampling interval and definition. Round-trip measurement and one-way packet delay are not the same metric.

Application steps

  1. Measure idle and load latency separately.
  2. Examine the average as well as high percentiles and jumps over time.
  3. Repeat the same scenario without mixing Wi-Fi, Ethernet and VPN conditions.

Practical example

If most of the samples are 20 ms, some 300 ms, the average may seem acceptable, but the call may feel stuck. Checking if these jumps occur at the same time as a large file upload will narrow down the cause.

Interpret the result correctly

A single universal good jitter limit does not fit every application. Vocoder, buffering and implementation tolerance are different. Match the measurement to the symptom experienced by the end user; Do not replace hardware based solely on a colored quality label.

Source and follow-up reading

Protocol or command details: RFC 3393. The steps and example scenario are IPScans editorial narrative.