VoIP call quality (MOS) calculator
Enter the latency, jitter and packet loss of a connection to estimate how a call over it will sound.
Callers are satisfied. Nothing is hurting quality much. This is about as good as a narrowband call gets.
One-way delay used: 100 ms (latency + jitter buffer + codec). Delay costs 2.4 R-points and packet loss costs 1.9.
How the estimate is made
The calculator uses a simplified version of the ITU-T E-model (G.107), the standard way to predict call quality from network measurements. It starts from a best-case R-factor of 93.2 and subtracts the damage done by delay and by codec compression and packet loss. The R-factor is then converted into a MOS, the 1-to-5 “mean opinion score” a panel of listeners would give.
What the inputs mean
- Latency — one-way network delay, usually about half the ping time.
- Jitter — how much that delay varies. The calculator assumes a jitter buffer of about twice the jitter, which adds to the delay.
- Packet loss — the share of voice packets that never arrive. Loss is assumed to be random.
Reading the score
Above 4.3 is excellent, 4.0 to 4.3 good, 3.6 to 4.0 fair, 3.1 to 3.6 poor and below 3.1 bad. Narrowband codecs top out around 4.4, so a perfect 5 is not reachable.
Limits
This is a planning estimate, not a measurement. It covers the narrowband codecs listed. HD codecs such as G.722 and Opus score higher in real use than this model can show. Echo, background noise and poor handsets are not included. To size the connection itself, use the codec bandwidth calculator.
FAQ
What is a good MOS score for VoIP?
4.0 to 4.4 is good: callers are satisfied. Below about 3.6 some callers start to complain, and below 3.1 most do.
What usually causes poor VoIP quality?
In most offices it is packet loss from congestion or weak Wi-Fi, followed by high latency. Voice tolerates loss and delay far less well than web traffic does.
Is this the same MOS my provider shows?
Not exactly. Providers usually estimate MOS from live RTCP statistics or run PESQ or POLQA measurements. Treat this tool as a planning estimate, not a replacement for monitoring.