Different Strategies for Dynamic Media Pipelines in Recording Video Calls
summary
Three practical pipelines: use the MediaRecorder API in-browser for quick client-side clips but risk lost recordings and high RAM as duration grows; run headless Chrome + extension on cloud VMs to avoid client failures and capture screens but pay heavy VM cost (roughly 4 cores and 8 GB RAM for a 720p recorder) and manage a pooled worker farm with segment uploads; or ingest RTP from the SFU and do GStreamer per-stream processing (or FFmpeg) to record separate tracks at near-zero extra CPU, scale per stream, and transcode reliably.
AI-written summary · about
chapters (7)
- 0:00Introduction to media pipelines
- 2:31Client-side recording approach
- 6:45Cloud browser recording
- 9:16Custom transcoder pipeline
- 12:46Stream-based recording
- 15:59Choosing the right approach
- 17:24Q&A session
related talks
12mGoing Live from your Browser without WebRTC
07mGoing Live from a Browser… with Another Browser
32mWebRTC Update
17mThe future of communication on the web
38mReal time front-end alchemy
24mVideo streaming on the web: Live and recorded video streaming
26m#javascript Audio/Video Streaming: #WebRTC APIs 101 explained
20mBeyond P2P: Video Routing in WebRTC