RTEnnui is a replacement for the media component of WebRTC. Its goals are portability, low latency, and control.
WebRTC's support for client-to-client data connections is good, but its media support suffers from design by committee. It's extremely opaque, has unpredictable latency and performance, and has requirements that are quite onerous in terms of patents.
RTEnnui fixes this by doing the media component in software, and using WebRTC only for the peer-to-peer connection. This gives us direct control over things like buffering and encoding.
You can find information on the design in DESIGN.md. The server component is at https://github.com/ennuicastr/rtennui-server .
RTEnnui is usable for both video and audio on Chrome, Chromium-based browsers, Safari, and Firefox. Video on Firefox is noticeably degraded, and if a Firefox-based browser connects to a session, video will be degraded for all users. This is because Firefox does not support WebCodecs (yet?), so it has to do its decoding in software.
There is a reason that this is a 0-dot release of RTEnnui. It's good enough that its author uses it in real-life situations, but it's far from perfect.
The API is documented with TypeDoc at
https://ennuicastr.github.io/doc/rtennui/modules/main.html . Note that the
main module (linked) is exposed when you import or require RTEnnui, or you
can include rtennui.min.js as a script, in which case the main module is
exposed as RTEnnui.
RTEnnui requires that you load libav.js
with support for at least demuxing MOV/MP4 and Matroska/WebM files, and
en/decoding VP8, e.g., the webcodecs variant.
You can find a complete (if simple) example of using RTEnnui in the server
repository, in samples/basic-server.