Mean Opinion Scores (or “MOS” scores) are a way to measure the quality of audio, such as the quality of audio in telephone calls. MOS scores have been used for years as a way to standardize the answer to the question: “How was the quality of the phone call?”
Think about it — if you ask ten different people how the quality of a phone call was, you will probably get a number of different answers (like, “it was ok”, “not bad”, “good”, “fine”). These types of answers are not standardized and are not very helpful to really get a feel for how good (or bad) a phone call was.
MOS scores standardize this and provide both objective and subjective measures to normalize how we look at the quality of a call.
Mean opinion scores are a term that is defined in an International Telecommunications Union (“ITU”) standard (ITU-T P.10) as “The mean of opinion scores” or “the value on a predefined scale that a subject assigns to his opinion of the performance of the telephone transmission system used either for conversation or for listening to spoken material.”
While MOS scores were designed to be used in “old school” voice networks, it has been extended for use with SIP and VOIP infrastructure. The specifications set forth various calculations depending on the voice technology used (and depending on what codec is used if SIP / VOIP is the communications protocol used).
In general, the calculated score represents a level of quality from “poor” to “best” (and is generally represented by a numeric such as 4.4 (for a call of “best” quality”) or 2.9 (for a call of “poor” quality).
You can read more about the actual calculation of MOS scores here, but in general, most PBX and ACD systems automatically calculate the MOS scores for calls. For example, the FreePBX switch (based on freeSWITCH) publishes the MOS score for each call in the Call Detail Record (CDR) (see here).