Placeholder text for a post on measuring call quality in VoIP systems. It will start with what MOS actually represents — a modeled estimate of subjective quality, not a direct measurement — and how R-factor feeds into it.
A middle section will cover the usual suspects behind a dropping MOS score: jitter buffer behavior, packet loss bursts versus steady loss, and codec choice.
The last part will probably discuss the gap between per-call MOS estimates and what callers actually report, and why both are worth tracking separately.