MOS – Mean Opinion Score
Mean Opinion Score (MOS) is a measure of voice quality that has been used in telephony for decades to assess human users’ opinions of call quality. The test is widely used in Voice over IP (VoIP) networks to ensure quality voice transmission, test for quality issues, and provide a metric by which to measure voice degradation and performance. MOS scoring is essential to ensuring both client and customer satisfaction. The measurement assigns a score from 1 to 5 to indicate VoIP call quality after it is transmitted and compressed using CODECs. The result can be useful in predicting VoIP call quality.
For instance, if a VoIP Service Provider’s average MOS score is 3.5, you can expect half of all users, and your customers to complain about poor voice quality. Typical VoIP calls fall in the 3.5–4.2 MOS range. A 5.0 score is generally unheard of, as it is considered perfect and has been compared to in-person, face-to-face conversations (though anyone using HD voice G.729 or OPUS CODEC) would argue that their MOS score must be a 5.0 – it’s that amazing!
The following chart provides MOS comparison values and can be used as a simple tool for comparing overall voice quality.
|5||Excellent||No effort is required to understand the language.|
|4||Good||Through attentive listening, speech can be heard without effort.|
|3||Fair||The language can be heard with a slight effort.|
|2||Poor||It takes a lot of concentration and effort to understand the transmitted language.|
|1||Bad||Despite great efforts, one cannot communicate.|
How is MOS determined?
By its very name “Opinion,” the MOS, Mean Opinion Score, was a subjective measurement used to test the listener’s perception of the voice quality and clarity of the communication. The test was performed in a ‘quiet room’ meeting specific size and noise requirements, in which listeners would receive and score calls on the quality as they perceived it. VoIP MOS score measurements, on the other hand, is usually more objective with the score providing a quality measure of the network over which it is carried.
MOS testing for VoIP phone networks is defined in the ITU-T PESQ P.862 standard, and is algorithm based rather than relying on live subjects’ opinions. The test is modeled on the subjective tests that were commonly used previously, assessing the voice quality of human beings using true voice samples as test signals. For accurate results, it is vital that modern telecom equipment be loaded with speech-like signals that are optimized in order to avoid unpredictable, or unreliable results.
What affects a VoIP MOS score?
As MOS is a relative scale, there are many factors that can affect voice quality such as bandwidth, CODEC, hardware, jitter, latency, and packet loss. Following are some examples:
The time it takes for a digital signal to travel from start to finish over the entire network, including traveling through routers, switches, and firewalls. The greater the distance, the greater the propagation delay.
The time it takes to digitize the signal based on the CODEC used as it’s sent over the Internet and decoded at the other end. More compressed CODECs such as G.729 have a higher packetization delay than a non-compressed CODEC such as G.711.
Jitter is the variance of the runtime of data packets. This effect is particularly annoying with Internet telephony as packets may arrive too late or too early to be output in time. The effect is reduced by a so-called jitter buffer, a special “data buffer”, but at the price of additional runtime, which is particularly disturbing in dialog applications.
The Callicity℠ Difference
Understanding that there is no better voice of the customer than the customer’s voice, you need to actually hear your customer and they you. Our managed PBX services use the latest, most advanced, servers, SIP trunks, and CODECs to ensure the best possible High Definition voice quality. Our phones and peripherals are designed with voice quality and call clarity in mind.
Call, Chat, or E-mail us today, and we’ll show you how to best hear the Voice of the Customer.