A “MOS” score refers to a “Mean Opinion Score”, and is a call center term that refers to a scoring system that is used to measure the quality of a voice communication (such as a phone call between a call center agent and a customer).
Simple Explanation of MOS Scores
In simple terms, a high MOS score (e.g., a 4 or a 5 on the MOS scale) refers to a very clear phone call (where both parties could hear each other well).
A low MOS score (such as a 1 or a 2 on the MOS scale) refers to a call that was choppy, or one with background noise where one (or both) parties had a difficult time hearing each other on the call.
Call center managers can use MOS scores to figure out if there's a problem with their phone system. If the MOS scores are consistently low, it's a sign that the call quality isn't good and there could be a connection issue. That could be a problem with the phone lines, the internet connection, or even the hardware like the phones or headsets the agents are using.
Technical Explanation of MOS Scores
For those (slightly) more technically inclined… a Mean Opinion Score (MOS) is a numerical measure used in digital communications and networking to assess the human-perceived quality of experience (QoE). MOS provides a numerical indication of the perceived quality from 1 (worst) to 5 (best).
MOS scores are typically calculated by gathering opinions from a panel of human listeners who rate the quality of test sentences read aloud over the communication system. Listeners rate the quality on a scale of 1 to 5, and the scores are averaged to get the MOS.
For practical purposes, and because gathering a panel of human listeners is time-consuming and expensive, a number of algorithms have been developed to estimate MOS scores. These algorithms take into account various factors that influence the perceived quality of a call.
For example, in VoIP (Voice over IP) telephony, factors can include:
- Latency: The time it takes for the voice data to travel from the speaker to the listener. High latency can cause delays in conversation and lower the MOS score.
- Packet loss: In digital networks, voice data is divided into small packets for transmission. If some of these packets get lost along the way, it can cause gaps or distortions in the received audio, lowering the MOS score.
- Jitter: The variation in the delay of received packets. If packets arrive at uneven intervals, it can cause audio to break up or sound jittery, which also lowers the MOS score.
- Codec quality: Different algorithms for compressing and decompressing voice data (codecs) can affect audio quality and hence the MOS score.
These parameters can be measured and fed into a predictive model (like the E-model for telephony) to estimate the MOS score. These models can provide a good approximation of the MOS score without needing a panel of human listeners for every call. However, they can't capture all the nuances of human perception, so actual listener ratings are still considered the most reliable measure of audio quality.
Determining a MOS Score
Many enterprise and open source telephony platforms calculate an estimated (or “objective”) MOS score. For example, the open source FreeSWITCH telephony switch provides a calculated MOS score in the call detail records of each call.
You can find more information about MOS scores (including the official definitions) at the International Telecommunications Union website and their standards (for example, check out ITU-T P.10/G.100 “Telephone Transmission Quality” standard recommendation here.