Hacker News Reader

Comment by Arathorn | Hacker News Reader

The current video rooms implementation is powered by Jitsi, which optionally exposes E2EE for clients which support it (Element Web on Chrome and possibly Safari) via insertable streams and megolm crypto from Matrix. It isn't hooked into Matrix's key identity system though.

In the nearish future we'll swap it to use Element Call which now has native Matrix E2EE over native Matrix calling and works on all platforms: https://element.io/blog/element-call-beta-2-encryption-spati...

In terms of crypto UX being problematic in Element as a whole: yes, there is still work to be done to improve the UX, and we're very aware of it.

View on HN

I've never heard of this and will be checking it out.

Have you cracked positional audio, ie letting people in the back sidebar convos (duck the main chat) or move around to have private convos?

Gaming has pulled this off, DayZ star citizen etc.

I've kicked around the idea of using some of these game worlds because they have positional audio and handle crowds way better.

It's the chat around the Watercooler that is the missing gap imo

Yes, we already have spatial audio working in Element Call (see https://element.io/blog/element-call-beta-2-encryption-spati...) - so this will come automatically when video/voice rooms in element start using Element Call.

Nioce!

We're also working on the Third Room project which has virtual worlds with 3D positional audio built on top of the same tech as Element Call. It's still very early, but the project is up here: https://github.com/matrix-org/thirdroom

I have one question about this. I am self hosting element and i have set up jitsi for group calls and this feature. Works great but unfortunately my kiddo instance is public.. is there any way to secure it easily or make it only available to element?

Is that Jitsi closer to the new Jitsi Meet or the old Jitsi Jabber/XMPP ?

It’s an embedded version of Jitsi Meet.

What the assumed time frame for the switch?

And once it happens, will it be a standard feature of synapse to deploy your own voice/video server?

I'm guessing it's about 1-2 months away. The hard bit is that we need to effectively have Element Call share the same matrix client as Element itself does, in order to embed Call into Element effectively. We call this matryoshka embedding, and it's looking promising to get it working in the next week or two, and then will need to actually update all three Element clients (Web, Android & iOS) to support it.

Obviously folks can also just click into a standalone Element Call to make it work (plus Hydrogen is also adding native Matrix calling).

Element Call doesn't currently need any custom server - it just uses your existing homeserver; Synapse, Dendrite, Conduit etc. However, it only scales up to 7-8 users - to go beyond that we'll need to add a SFU, which we're also working on.

A MCU or a virtual apex could be better in order to ensure everyone hears the same thing. And a virtual apex combined with large events could also provide a decentralised persistence layer.

We're going for a hybrid SFU/MCU architecture - c.f. https://2021.commcon.xyz/talks/extending-matrix-s-e2ee-calls...

Thanks.

And thanks for all the great work as well.

Any MSCs for the RTC part of the Element Call?

Yup, MSC3401 https://github.com/matrix-org/matrix-doc/pull/3401

Reply on news.ycombinator.com