MOSSkalen

MOSSkalen, or MOS scales, are subjective rating scales used to quantify the perceived quality of multimedia content as judged by human observers. They revolve around the concept of a mean opinion score (MOS), a numerical value that summarizes a listener’s or viewer’s overall impression of quality.

The most common MOS scale is a five-point scale, with anchors typically defined as 5 = excellent, 4 = good, 3 = fair, 2 = poor, and 1 = bad. Some experiments use alternative scales (for example 1–4) or provide a continuous presentation (such as a 0–100 scale) by rescaling the ratings. MOS values are averaged across raters to produce an MOS for each stimulus, stimulus type, or condition. When comparing degradations to a reference, researchers may use DMOS (difference MOS), which expresses the perceived drop relative to the reference.

MOSSkalen are applied in areas such as speech and audio quality assessment, video quality assessment, and broader multimedia QoE studies. Standardized methodologies often involve controlled subjective tests where participants rate stimuli under defined listening or viewing conditions, with sample sizes and procedures designed to ensure reliability and repeatability. Results from MOSSkalen are used to guide encoding decisions, streaming configurations, and quality improvements in telecommunications and multimedia services.

Limitations include subjectivity, cultural and individual differences, and potential biases in test design. Proper pilot testing, representative raters, and transparent scoring procedures are essential for reliable MOS-based evaluations. See also: mean opinion score, DMOS, subjective quality assessment, ITU recommendations.