WebRTC Metrics
A comprehensive overview of WebRTC statistics, derived indicators, and observable signals, to better understand call quality, connectivity, and user experience in rtcStats
mos
The Mean Opinion Score (MOS) of an audio or a video stream.
Description
Real number; Between 1 to 5
MOS is a metric that represents the quality of audio or video communication as perceived by the user. It is traditionally measured by having users rate the quality of a call on a scale of 1 to 5.
This subjective feedback is then averaged across multiple users to obtain a mean opinion score.
rtcStats calculates MOS for both audio and video:
- For audio, we use either ITU-T G107.x or our own advanced model. In both cases we take into account the codec type
- For video, we rely on bPPPF model or use our own advanced model
Interpreting Values
| Range | Description | |-------|-------------| | 4.3-5.0 | Excellent (toll quality). Users perceive no issues | | 3.5-4.3 | Good. Most users are satisfied | | 3.0-3.5 | Fair. Some users are dissatisfied | | 2.5-3.0 | Poor. Many users are dissatisfied | | <2.5 | Bad. Nearly all users are dissatisfied |
For reference, traditional phone calls typically score 3.5-4.0.
User Experience Impact
- MOS directly maps to perceived quality. A score drop from 4.0 to 3.0 is the difference between "sounds fine" and "hard to understand"
- Below 2.5, users will likely abandon the call
- Small drops (e.g., 4.2 to 3.8) are usually tolerable. Drops below 3.0 trigger complaints
Troubleshooting
- MOS is derived from jitter, RTT and packet loss. Identify which input metric is dragging the score down
- Fix the worst offender first - usually packet loss has the biggest impact, then RTT, then jitter
- Check the individual contributing metrics to pinpoint the root cause before making changes
See also
Notes
- MOS is calculated independently by rtcStats and is not part of the metrics that WebRTC offers out of the box. The calculation is based on audio related metrics such as jitter, RTT and packet loss
