Introduction to WebRTC Statistics and getStats() API
summary
Practical walkthrough of using getStats shows how to poll every second (avoid calls under ~150 ms) to read outbound-rtp RTT, accumulate samples, divide by two for one‑way delay, and feed a simple E-model where <250 ms is good and >500 ms is bad. Examples show resolution shifts from 1080p to 640p, throughput spiking to ~2.6 Mbps and drops with ~40% packet loss, and an eight‑minute disruption window; user actions (turning off video) often correlate with self‑diagnosis.
AI-written summary · about
chapters (11)
- 0:00Introduction to Callstats
- 0:48Multimedia Protocol Basics
- 3:01RTCP and Network Feedback
- 4:19Measuring User Experience
- 6:54Call Metrics and Tracking
- 10:21Data Collection Levels
- 12:06Using the getStats API
- 14:41Analyzing Call Pipeline Data
- 18:38Audio Quality Modeling
- 21:33Visualizing User Disruption
- 24:26Conclusion and Resources
related talks
27mWebRTC: Real-Time Communications
21mOptimizing the customer experience
21mDecisions & considerations in building your WebRTC app
21mDemystifying WebRTC
37mWebRTC: From The Ground Up
31mWebRTC: Real-time communication without plugins
56mYou're on Mute! WebRTC and our Lives on Screen
33mWebRTC Change Communcations Forever