Reference
MOS Score Explorer
Select a codec, slide the score, and see what each MOS level means for real calls.
G.711 is the reference codec. Its only impairment comes from network conditions.
- ›Conversation flows naturally with no perceptible issues.
- ›Comparable to a landline call (narrowband) or better (wideband).
Typical network conditions at this score
These are approximate values derived from the ITU-T G.107 E-model. Network impairments combine nonlinearly; a single number cannot fully explain a given MOS.
Packet Loss (random)
0%
Approximate random loss that would produce this MOS at zero delay
One-Way Delay
<5ms
Approximate mouth-to-ear delay that would produce this MOS at zero loss
Jitter Buffer Impact
Jitter itself does not reduce MOS directly. It is the jitter buffer discard rate (packets arriving too late to be played) that degrades the score. High jitter forces a larger buffer, adding latency, or a smaller buffer, increasing effective packet loss.
Frequently asked questions
What is a good MOS score for VoIP?+
A MOS score of 4.0 or above is considered good. Most callers will not notice any quality issues. Scores between 3.5 and 4.0 are acceptable. Below 3.5, callers start reporting problems.
Why do different codecs have different maximum MOS scores?+
Each codec introduces a baseline impairment from its compression algorithm. G.711 uses no compression and achieves the highest narrowband score (4.41). Compressed codecs like G.729 start lower (3.92) because the compression itself reduces quality even before any network impairment.
Does the MOS score account for network conditions?+
Yes. The E-model (ITU-T G.107) calculates MOS from the codec's baseline impairment plus degradation from packet loss, latency, and jitter. The MOS Explorer shows how network conditions map to each score level.
How is MOS calculated?+
Modern MOS is calculated using the ITU-T G.107 E-model, which produces an R-factor (0-100) from codec type, packet loss, delay, and jitter. The R-factor is converted to MOS using a standard formula. This replaced the original method of human listening panels (ITU-T P.800).