LIVEKIT BLOG | Link

Using a transformer to improve end of turn detection

LiveKit’s new End of Utterance (EOU) model tackles voice AI’s turn detection problem by combining semantic analysis with voice activity detection (VAD), cutting unintentional interruptions by 85%. The open-source transformer model analyzes transcribed speech in real-time to dynamically adjust silence timeouts, avoiding premature responses during natural pauses. Integrated as a plugin for LiveKit Agents, it currently supports English, with multilingual support planned in the future.

JITSI BLOG | Link

AV1 and more … how does Jitsi Meet pick video codecs?

Jitsi Meet now uses AV1 as its default video codec, offering 30-50% better compression than VP9/H.264 for high-quality streaming in low-bandwidth scenarios. Despite higher CPU demands, Jitsi added adaptive codec switching and resolution controls for older devices, while leveraging AV1’s Dependency Descriptor header to enable scalable video routing. The update also introduced full SVC support for efficient single-stream encoding.

FAL AI GITHUB | Link

New Open Source AI-Powered Video Editor by Fal.ai

Fal.ai has introduced a lightweight, open-source video editor that can run entirely in your browser. It’s built on top of Next.js, ffmpeg, and Remotion and integrated with the Fal.ai APIs. It features generating audio and video assets, real-time previews, and custom file uploads.

Whether you’re a developer exploring new tools or looking to create custom video solutions, this project is free to try and fully open source.

TAILSCALE BLOG | Link

How NAT traversal works

Let me share my favorite article on NAT traversal and NATs in general. Although NATs have solved one of the biggest challenges - a limited address pool of IPv4-based networks, they remain an issue for anyone who tries any P2P technology, like WebRTC.

This absolute encyclopedia of an article will tell you everything about any technique that can crush straight through NATs.

Some time ago, Samrat Man Singh shared an X post demonstrating his project – a livebook allowing to talk to Google Gemini, build with Membrane WebRTC. We were curious about how did he do it, so we asked Samrat if he could do a deep-dive – and here it is!

If you're working with WebRTC and LLMs, you’ll want to check this out.

We teamed up with L7MP and created a simple, globally distributed streaming service based on Kubernetes, Stunner and Elixir WebRTC.

Check out the video below & check how your connection quality changes depending on a cluster you are connected to and network conditions.

Missing RTC.ON already? Us too, so we prepared a little something to get together before September :)

Join us for the first-ever edition of Kraków Video Tech Meetup – a place for all devs interested in WebRTC, streaming & computer vision!

🕕 WHEN: Feb 6, 6 PM CET

📍 WHERE: online (link will be added to our Meetup) & offline in our office!

For the first edition, we'll be joined by 3 audio & video tech experts, ready to share their knowledge with us.

Check out the full lineup:

We hope you enjoy the RTC.ON newsletter as much as we do. It's great to see all of you joining us each month for a little multimedia walk through – thanks!

Here are some more ways to connect with us:

Want to share RTC.ON newsletter with a friend? Here is a link to our sign up page, including the archive of all past issues.