What is MOS (and why is important for you!)

Have you ever suffered a very bad quality experience when you were watching a film on television? Perhaps because the image was so pixelated sometimes. Perhaps because the sound was so dirty, or it was not synchronized with video.

Multimedia platforms and other video providers are aware of the relevance of the quality in this experience. They could know if they are satisfying viewer expectation thanks to MOS application.

MOS: Why you should be aware about this technology

MOS has been the most common metric used to measure the overall voice call quality for decades. In fact, it has been standardized by the International Telecommunications Union (ITU-T). MOS Score refers to a numerical measure of the human-judged overall quality of voice and video sessions. MOS is referred to Mean Opinion Score. As a metric, it is defined as a number, referring to the human-judged overall quality of an event or experience. MOS is a key indicator of the quality of voice and video sessions, especially in media industry.

MOS is usually defined as a numeric value. It is ranging from 1 to 5, where 1 is a Bad quality, 2 for Poor, 3 measure Fair experience, 4 is set for Good and 5 for Excellent, according to the user experience.

Due to the human tendency to avoid perfect ratings (now reflected in the objective approximations), somewhere around 4.3 – 4.5 is considered an excellent quality target. On the low end, video content quality becomes unacceptable below a MOS of roughly 3.5.

Mean Opinion Scores are the average of several other human-scored individual parameters. Because of that, it could be applied in all the scenario where human subjective experience and opinion is relevant. Anyway, the current case use is related to judge digital and broadcasting experiences.

For example, MOS is applied to measure static image compression (e.g., JPG, GIF), audio codecs (e.g., MP3, Vorbis, AAC) and video codecs (e.g., H.264, Opus, VP8). It is also very commonly employed in streaming sessions where network effects can degrade communications quality.

MOS evolution

Originally, Mean Opinion Scores were derived from surveys of expert observers, but nowadays, and thanks to technology development, MOS is often produced by an Objective Measurement Method emulating a human ranking.

Before to be applied, it is necessary to originate polls and test about video contents. Several current standards can be traced back to expert listeners and observers in distraction free quiet rooms subjectively logging experience scores. A MOS itself is a metascore, averaged from several individual components of session quality. A metascore is the rating of something. Scores are assigned to reviews and weighted average are applied to summarize the opinions range. The result is shown in single number that captures the essence of reviews in one metascore

In other words, in a MOS test, the persons listen or watch to short samples, where every piece consists of two to five sentences. The total MOS score is then the mean of all individual results.

It is important to note that audio and video communications isn’t scored by a panel of individuals. Today, they are measured by algorithms (Objective Measurement Methods) that try to be equal to human experience. Video MOS is the first company able to emulate this measure monitoring the video experience as if human will be doing it thanks to IA algorithms developed by its team.

These algorithms should submit the standards ITU-T’s P.800.1 (regarding objective and subjective scoring of telephone transmission quality) and P.863 and J.247 (covering speech and video quality, respectively).

The standard defines how to calculate MOS Score based on multiple factors such as the specific codec used.


No Comments

Post A Comment