Why is self-hosted voice chat so hard?
from early_riser@lemmy.world to selfhosted@lemmy.world on 13 Feb 02:32
https://lemmy.world/post/43072565
from early_riser@lemmy.world to selfhosted@lemmy.world on 13 Feb 02:32
https://lemmy.world/post/43072565
All this talk about Discord replacements plus my own experience attempting to host a Synapse has got me wondering why it seems so hard to implement voice chat.
Stupid idea: back in 2022 I got an Asterisk server working on a raspberry pi over AREDN without too much trouble. What’s stopping people from just using a PBX like that for voice chat?
threaded - newest
Isn’t TeamSpeak still a thing?
Yeah, from what I understand, standing up a Teamspeak server is pretty straightforward.
I think they just announced a big new version just in time for Discord to tell us all to fuck off
I thought it was only voice though. Not screen share or chat.
Yeah, it has chat. Nothing too fancy, at least not back when I last used it (which, granted, is decades ago) but chat it has.
It does have screen sharing now! The Linux client currently can’t share system audio, but I think they’re working on that
Not FOSS or open source in any sense. You could still say it’s self-hosted, but I suspect most people self-hosting care about this.
Nothing is stopping it, it’s just not particularly convenient because it’s designed around the limitations of the phone system.
SIP could handle it all if you wanted though.
It’s not, but the people who are asking are often not tech-savvy, and any amount of self-hosting will be hard for them
Try mumble if you just need voice. Just fire up a docker container and open a tcp and a udp port. The settings are under-documented so things like auth are tough to set up.
I second mumble, it’s a 5 min job to fire it up and default servers settings are enough to get going out of the box.
Simple 1:1 audio stream is easy.
Groups, screen sharing, noise canceling, NAT traversal, mobile apps, and all those extra features people have come to expect are hard.
Exactly!
people act entitled as if all that you mention was trivial and that somehow FOSS devs “owe” people, but we only see those big corpos make it happen because… well, they’re big corpos, burning VC money on makint it happen and making it happen in a controlled jail.
I have honestly not seen anyone acting like they are “owed” these things by FOSS developers. We just want them.
I have seen lots of people. Mostly not here, but that’s because we here know better (I’d hope). Runs along with usual complaints such that they can’t move from a platform with 9trillion captive users to a new budding platform, conveniently forgetting that when they began Shitter and stuff also had like 0 users yet people did move.
Groups: just simple Chanels are fine, password lock them if you want.
Screen sharing: one at a time should be fine. Self hoster can configure max bit rates.
Mobile apps: building your app to be multiplatform is a lot easier than it was a decade ago.
Mumble will do all of that except screen sharing. Only the server has to deal with NAT.
As for “why is it hard to self-host”, it is only NAT traversal.
TURN, STUN, ICE, etc. are not fun to debug. Not sure if anyone still bothers fiddling with TOS/DSCP on their router. You can build a voice server that just exposes a TCP port, but… latency. And corporate firewalls love to randomly block some UDP port ranges but not others.
The one point that has basically been solved is NAT traversal. Thanks to Wire guard, Tailscale and the like. The relevant parts are open source and can be used basically as a library.
VDO Ninja is really nice. My friend self hosts it, and it didn’t seam that hard.
So as mentioned we have both Mumble and Team Speak if you are looking for a self hosted VC.
Acronyms, initialisms, abbreviations, contractions, and other phrases which expand to something larger, that I’ve seen in this thread:
3 acronyms in this thread; the most compressed thread commented on today has 9 acronyms.
[Thread #88 for this comm, first seen 13th Feb 2026, 05:10] [FAQ] [Full list] [Contact] [Source code]
Good bot
Use mumble
It’s easy. Mumble. Or the thing you used probably still works.
But you see, people never actually seek a discord alternative. They want a discord alternative that includes all the features in one app that is also federated, AND end to end encrypted, and each one makes things vastly more technically challenging and resource intensive and then you want them together.
A little secret: Matrix is much, much easier to host if you disable encryption and federation. Federation to many servers is the main performance killer, and “failed to decrypt message” will all disappear if you disable encryption.
If you go with anything using livekit (stoat/revolt , lasuite meet), voice is not very hard per se (just a bunch of udp ports required).
It’s video that will get your CPU to its knees
I just can’t get over the name change from Revolt to Stoat, but at least a stoat is cute AF.
Cease and desisted, but yeah it’s sad (and annoying code-wise because you have to rename everything)
Don’t forget about teamspeak!
This has big XKCD Energy. It almost feels like an exact recreation of the comic but with tech:
<img alt="" src="https://lemmy.world/pictrs/image/1d19dc6c-c42c-4c61-a8e6-3e9a6253272c.png">
I meant the OP more as a lament about it being hard rather than a quip about it being easy.
Though upon reflection it’s not the voice chat that’s a problem, it’s the fact that Discord is a lot of things, a chatroom, a VOIP service, and so on, and recreating all those things on top bolting on federation (which I don’t see as a desirable feature in this case) is what makes it so hard.
Since no one has mentioned it yet -
If you haven’t tried setting Synapse up with the ansible playbook you should. It’s almost push-button and has 1:1 voice calls by default. Setting up group voice from there is a bit challenging, but the playbook has a section for it as well if you’re willing to try.
Edit: cookbook - > playbook
I haven’t messed with Ansible in a bit. Do you mean playbook? Or is cookbook some other concept I haven’t learned about yet?
Nope, youre right, playbook is the right term. Got it mixed up with chef.
It’s not! Use SonoBus; it’s dead simple, and superior to Discord. It’s far lower latency, with customizable filters, peer-to-peer; and totally free.
Now if you want emojis and video and rambling channels and stuff, you will have to go elsewhere.
What a clever name!
Update: I got Mumble working without a lot of grief. Their mobile client isn’t great though. I might try Stoat.
Federation just complicates things, as it’s just for a myself and a few friends.
f-droid.org/packages/se.lublin.mumla is not so bad as a mobile client for Mumble.
I’m sure it is but I have an iphone