Video Freeze Events

Understand video freeze event counting — how CallMeter detects frozen video, thresholds, and what freeze events mean for the viewer experience.

Video Freeze Events counts the number of times the video froze during a call. A freeze occurs when the video decoder cannot render new frames, causing the last successfully decoded frame to remain stuck on screen. Each freeze event represents a distinct moment where the viewer saw a motionless, stuck image instead of live video.

This is the single most user-visible video quality problem. A viewer may tolerate lower resolution or frame rate drops, but a frozen picture is immediately obvious and disruptive.

How It Works

CallMeter monitors the video decoder's frame output. When the decoder stops producing new frames for a detectable period, a freeze event is recorded. The freeze continues until the decoder receives enough data (typically a keyframe) to resume rendering.

Each freeze event is counted once, regardless of duration. A half-second freeze and a five-second freeze both count as one event. For duration measurement, see Video Freeze Duration.

Distinct from low FPS

A freeze is not the same as low frame rate. At 10 fps, the video is choppy but still moving. During a freeze, the video stops completely — the same frame stays on screen until recovery. CallMeter distinguishes between these two conditions.

Why It Matters

Video freezes have an outsized impact on perceived quality. Research consistently shows that users rate a call with freezes significantly worse than a call with uniformly lower quality. A single freeze event breaks the sense of real-time presence that video calls depend on.

For enterprise VoIP testing, freeze count is a critical pass/fail metric:

Load testing — Freeze count at scale reveals the capacity ceiling of your video infrastructure
Network qualification — Freezes during baseline testing indicate the network is not suitable for video
Codec evaluation — Comparing freeze counts across codec configurations helps select the most resilient option
Regression testing — Any increase in freeze count after infrastructure changes indicates a regression

Thresholds

Level	Freeze Count	Interpretation
Good	0	No freezes, smooth video throughout
Warning	1 - 3	Occasional freezes, noticeable but tolerable
Critical	Above 3	Frequent freezes, severely degraded experience

Common Causes of Video Freezes

Cause	Explanation
Burst packet loss	Multiple consecutive video packets lost, making frames undecodable
Keyframe loss	A keyframe is lost, and the decoder cannot recover until the next one
Network congestion spike	Sudden bandwidth reduction causing the decoder to starve
Jitter buffer underflow	Buffer runs empty, no frames available to render
CPU overload on decoder	Decoder cannot process frames fast enough to maintain output

How to Fix It

Identify the cause — Correlate freeze timing with Packet Loss Rate, Video Bitrate, and Jitter to determine what triggered each freeze.
Enable NACK retransmission — NACK-based recovery can recover lost packets before they cause a freeze.
Shorten keyframe interval — Reduce the Keyframe Interval so recovery from corruption happens faster, limiting freeze duration.
Increase bandwidth headroom — Video needs consistent bandwidth, not just average capacity. Add 30% headroom above the target bitrate.
Improve QoS — Apply DSCP marking and traffic shaping to protect video from competing traffic.

Video Freeze Duration — Total frozen time; the severity companion to freeze count
Keyframe Requests Sent — Requests triggered by or during freeze events
Keyframe Interval — Maximum recovery time between keyframes
Video FPS — FPS drops to zero during freeze events
Packet Loss Rate — Network loss is the most common freeze trigger