MOS calculator
Network tests give you three numbers — delay, jitter, and packet loss — and none of them answer the only question that matters: will the call sound acceptable? This turns those numbers into a call quality score, using the same industry standard we apply to every call we measure.
How long your voice takes to reach the other person. Halve a ping/round-trip figure to get this.
How uneven the arrival timing is. Under 30 ms is fine; past that, calls start to warble.
Share of audio that never arrives. Under 1% is inaudible; past 3% words go missing.
The default on nearly every business phone system. Best quality, most bandwidth.
Indistinguishable from a good landline. Nobody will comment on the call quality.
- R-factor
- 89.7
- Effective delay
- 70 ms
How to read the score
| MOS | Verdict | What callers experience |
|---|---|---|
| 4.3 – 4.4 | Excellent | Indistinguishable from a good landline |
| 4.0 – 4.3 | Good | Clear and professional; nobody comments |
| 3.6 – 4.0 | Fair | Usable, but something sounds off |
| 3.1 – 3.6 | Poor | Callers complain; words get lost |
| Below 3.1 | Unusable | Most people hang up and call back |
The arithmetic, openly
This calculator implements the ITU-T G.107 E-model. It builds an R-factor from a base of 93.2, subtracts an impairment for delay — with a sharp penalty once effective delay passes 177 ms, the point where people begin talking over each other — and subtracts a second impairment for the codec and packet loss, weighted by how gracefully that codec degrades. The R-factor then maps onto the MOS scale.
The same code scores the call measurements we publish, which is deliberate: applying one public standard to everything means our numbers can be checked rather than trusted. Read the full methodology.
MOS questions
What is a MOS score?
Mean Opinion Score — a 1-to-5 rating of how good a phone call sounds. It began as literal human listening panels rating recordings, and the modern version calculates the same number from network conditions instead. For business calls, 4.0 and above is the target; below 3.6 people start complaining.
What is a good MOS score for VoIP?
4.0 or higher. The practical ceiling depends on the codec: G.711, which most business systems use by default, tops out around 4.4 — so 4.3+ is effectively perfect and anything at 4.0 or above sounds professional. Below 3.6, callers notice; below 3.1, they struggle.
How is MOS calculated from network conditions?
Through the ITU-T G.107 E-model. It converts delay, jitter, and packet loss into an R-factor between 0 and 100, applying a sharp penalty once effective delay passes about 177 ms (the point where people start talking over each other), then maps that R-factor onto the familiar MOS scale. This calculator implements that standard directly — the same code that scores the call measurements we publish.
Why does jitter count double in the delay figure?
Because a jitter buffer has to hold roughly two jitter-widths of audio to smooth out uneven arrivals, and that held audio is added delay. It is why jitter hurts twice: once through the buffer delay it forces, and again as discarded packets when it exceeds what the buffer can hold.
Where do I get the numbers to put in?
Our packet loss test measures all three from your browser in about ten seconds. If you already have a ping result, halve the round-trip time to get one-way delay. Provider dashboards and PBX call-detail records usually report jitter and loss per call as well.