Hey LiveKit Community!
I recently open-sourced a self-hosted live streaming and media management application built around LiveKit’s WebRTC infrastructure, and I wanted to share the repository with the community.
The project leverages LiveKit for low-latency guest studio rooms and real-time media transport, bridging WebRTC directly into production software (like OBS Studio) via WHIP and RTMP egress, while enabling multi-platform social media broadcasting.
If you’re building self-hosted interactive video tools or working with custom LiveKit egress/ingress pipelines, feel free to check out the repo or contribute!
GitHub Repo: GitHub - docisit/itg-media-engine · GitHub
Feedback, questions, and PRs are always welcome. Appreciate the awesome work the LiveKit team and community put into this stack! I am running this project differently on my production Server, and stripped out any leaks I may have. But, With Docker Containers you can spin this up, and try it out. On my production, I am running a mix or pm2 services, and docker containers.
It’s taken me a month to actually load this to GitHub. I’ve been wondering my best way to share it, without leaking my own info, and still making sure it will run for others.
We all know that feeling of when we finally get something working, and just keep thinking what can make it better. Rather than using it, I keep working on new things. However, I have done 3 recent interviews and had completed a couple before those, while beind an ATT Wi-Fi Router, for ISP connection. Which, really helped me tighten things up. And, caused havoc at first, for the WebRTC. Though, I still pulled it together.
I’m running on an older Server, so I have held up on my talking Avatar. Though, I was able to make it work, by pushing the work load to the Users device, and this allowed it to pull their systems Audio. My old CPU can not handle the Math for the talking avatar. Though, it handle my chat bubble, connected to open ai, qwen model.