Mumble: Open-Source, Low Latency, High Quality Voice Chat
mumble.info
mumble.info
Prior to Opus, Mumble used CELT.[2] This was a precursor to the Opus codec developed by Xiph.org, makers of the very well known Ogg container format and Vorbis audio codec.
Added noise suppression would also be great, especially a FOSS-based one (Discord uses AI-based noise suppression from Krisp [0] -- a Berkeley, CA based for-profit startup).
I'm curious if there are any FOSS community projects in this noise suppression space. Given the rapid rise in WFH culture post covid-19, I can see an explosion in demand for s/w products in this space.
Drives me up the walls when I mention discords awful quality and people just say "I don't know what you're talking about they have the same codec".
It's only a piece of the puzzle. Not to mention the constant robotting because some of their servers are melting or something.
Host your own voice comms :) Even the cheapest 2€ a month box can handle a decent sized party with no issues.
Convincing everyone else involved to switch over to it however is a near-insurmountable battle.
Many years ago I was playing on a friends Minecraft server, while in a then Skype call with some friends on the server
One user couldn't get Skype to work on their Arch Linux install, and proceeded to spend almost an hour badgering us and whining to switch to Mumble so they could "join the fun"
From our perspective, shifting our entire group conversation over to a brand new voice chat system just for one person seemed like an absolute joke. So they were left wanting
I love it, because I'm an engineer and know what all those buttons and sliders do. My friends? they didn't like it at all. Thus making the switchover even harder.
This issue has been mentioned in the past already, and I think Mumble devs acknowledged this in the past... having a well polished UI is so important for user adoption! Just throwing all technical choices to the user's face is rarely the right solution.
I used it with my friend in 2015 as a sort of galactic walkie-talkie on our phones -- he was in another country but we'd just walk around outside and talk to each other. It felt pretty futuristic at the time.
After a few minutes of continuous audio transmission though there would be a significant buildup of latency, so we developed the habit of saying "pineapple" to each other to measure the lag. Best record was 12 seconds.
Nice to see you had been having fun with it. Skype however had been there since a lot earlier, right? I guess that having to exolicitly "call" your friends removed a bit of that "walkie-talkie" feeling, but still...
And why is there no passthrough mode setting for voice channels?
Plenty of people out there with decent usb mics and a cheap little dac/amp setup.
I always get angry when companies use it for commentary on esports because its just that damn awful
It's just so bad in discord I don't know what they are doing
I know that a few open source games use it for in-game chat. Empty Epsilon [1], and my own Space Nerds in Space [2] both use it for this, and there are probably others.
[1] https://daid.github.io/EmptyEpsilon/ [2] https://smcameron.github.io/space-nerds-in-space/
I personally had it enabled (by accident) for over a week, and during that week I noticed people were either ignoring me or not hearing me. I discovered it was the latter when I disabled the noise suppression option. Krisp was suppressing my voice in addition to noise.
Other people I know who’ve tried it have had similar results, so I’m wondering if we’re using it wrong or if Krisp just isn’t worth using at this point.
The FSF even offers this as one of the few reasons you wouldn't use Copyleft. It actually would be better for society if everybody copy-pastes the Opus code and refuses to let you see their source code, than for them to use a proprietary codec instead.
> Speex was the first audio codec we used, until it was superseded by CELT in 2009 (Mumble 1.2.0), which in turn was superseded by Opus in 2011 (Mumble 1.2.4).
> We retained support for [Speex] for 10 years and we believe all active servers are currently using Opus.
Is this because the server didn't have a certificate from a trusted CA? In which case the fix is that Mumble could integrate ACME to get certificates from Let's Encrypt or whoever
Or is the situation that Mumble doesn't integrate the WebPKI and so it expects the user to make trust decisions for each certificate, which is pretty hostile ?
> That is an issue of general technical illiteracy
I guess that's kind of true, but I'm not sure I should need to understand the correct range of manifold pressure for the engine in a motor vehicle to operate it, for example. "Just do what is obviously the correct thing" seems reasonable in both cases.
In my experience, a lot of people who set up a Mumble instance don't have an actual domain name, so they can't get a CA certificate, only self-signed. Most people do set up at least a dynamic DNS of some sort. But as long as you're doing that, you might as well pay the extra $10/year to get a domain, in my opinion.
Most dynamic DNS providers got a default shared domain name added to the Public Suffice List e.g. dyndns.example might be on the PSL and then you can have your server be named etskinner.dyndns.example when you call their dynamic DNS service.
In this case Let's Encrypt is quite happy to give you a certificate for etskinner.dyndns.example since you control it. Unlike a web server, the Mumble server can't trivially bake the elements needed for this into its functionality, but it shouldn't have a hard time in the two easy cases:
1. There is no web server for this DNS name, spin up a temporary web server, answer Let's Encrypt queries until they give you a certificate, then spin it back down
2. This machine is the web server, so, have the user tell us how to pass http-01 challenges on that existing web server.
That doesn't cover every corner case, and it is one more notch on your "Duplicate certificate count" rate limit if you do have an HTTPS web site on the same name from Let's Encrypt, but I'd guess 95% of users who have a working Murmur and either a Dynamic DNS setup or their own "proper" DNS setup would get a working system and a further fraction would have some trivial problem they'd fix and after that it would Just Work™.
Why is Discord not a good choice in your opinion?
On mumble however, I start the client and it just works. Also, it instantly reconnects should the connection break.
Can take a bit of time, but needs to be non-clunky.
Unless Google is employing hackers to secretly integrate reCaptcha into other apps, it is coming from Discord no matter who creates the captchas
You can, but it will be painful with captcha and email submit links for your new IPs.
Please let me know if that's not the case.
Discord having the slack-like features, video streaming and social networking integrated in a nice UI makes it a better all purpose consummer product unfortunately.
I understand that connecting to hundreds of servers wasn't a thing in 2015 but it is just so high friction to talk to people about video games.
Well... no. The way one configures users and permissions in Mumble clearly didn't pass the mom test.
I have similar qualms with how users identify. "What? are you telling me that there are these weird files that are called <certificates> and that if I want to move between my PC, laptop, or phone, I need to hunt for this file and copy it into all my devices?" (obviously at this point people started logging in as "Joe2", "Joe3", etc. in my server)
People with a speakers setup sound awful on all software. Some kind of headphones or earbuds cost less than 10 quid.
The feedback cancelling wasn't the problem. It's a feature that should scarcely even exist.
The only time it's ever really good is when there is some sort of interactive one to many setup. A hands on demo from a lecturer where for some reason they can't wear some wireless buds.
I haven’t used it in the past year, but if I were to guess I’d say it’s in the tens of milliseconds, definitely not higher.
I use Steam voice chat to talk to my brother and I find it by far the best one. Low latency and no filter BS. I don't think it is P2P though.
I and my friends used Ventrilo earlier and the level of QoS you get from hosting on a low load computer with like 4ms ping to each node is insane compared to modern alternatives ...
For a two way call Ventrilo is essentially P2P.
I used mumble for a few years to bridge a VHF radio between an airport and a remote site via the internet. The tricky bit was getting the push-to-talk on the radio to work, which required some circuitry and some source code tweaks...
I can see FOSS software outperform Discord, if you pick an aspect, like "low latency audio". But if someone wants to make FOSS successful, they better focus on the user experience. People sacrifice a good bit of everything else, just to get a better UX, and the more developers realize this, the more potential their projects have to succeed.
You can build actual _communities_ on Discord, it's not just about the voice chat. It brings the people together in a way they can share rich content in one space.
If you're on voice chat you will very often want to interact with them by sending images, videos or links back and forth, and can do this seamlessly by using Ctrl-V.
This is why IRC pales in comparison to modern chat solutions, and proponents will start rambling about open protocols or whatever stuff the end user doesn't care about. Can you paste a screenshot into the chat? Thought so.
On the other hand, IRC could easily be just used as an IM backend to a hypothetical FOSS discord alternative. You would press CTRL+V in the client, it would upload your clipboard image to the server, return you an URL, and that would be pasted into the chat box, which would then embed the image. GitHub's editor does this for example and it's a smooth experience, and perfectly compatible with a text-only protocol.
It can be nice to see videos and images, and it can be annoying. Quite often videos and images dumb down the conversation as people post memes instead of saying anything meaningful.
It can also be handy to see the image right there and not have to click on a link to open it in your web browser to see it, but that just makes the experience a bit more seamless and convenient, and is not something that is going to make or break a community.
Plenty of communities have formed on IRC.. big and small, even without inline images and video.
The last version of BBB released a few weeks ago apparently reduced resource-usage by up to two thirds server-side and brought in long-awaited features. Or so i heard from friendly hosting collectives, who definitely recommend to give it a go.
A) it's free, as opposed to a vent or TS server, which while they are not expensive, it's still a barrier to setting one up.
B) the free tier has quite a lot of functionality without paying for servers, even for a lot of players (like a World of Warcraft guild).
C) it merged voice Comms with a community hub where people could communicate and share things relevant to their game (to use the WoW scenario again, raid organising, upcoming patch discussion, guides and other helpful information) in an organised and central location.
I recognise there is a bunch of issues with discord, I've had it have complete melt downs when the voice systems have broken, it can be a real resource hog, and don't get me started on the security and privacy (it's not great), however because of the additional functionality I still think it's a great bit of software.
I'm so glad I don't have to be in Warcraft guild Facebook groups anymore!
My purpose was not to suggest you were wrong, just present my experience and thoughts about why it's become so widely used _despite_ the issues you've raised.
I think people overcome the awkward UI because of the perceived benefits of the platform verses alternatives. That has certainly been my experience, which I appreciate is a single data point.
I'm sure you're right! The robustness of Discord's call quality definitely seems to be much better than average (though a lot of this comes down to people's local networking hardware).
It does work great for that.
My main problem with Discord is that I can't get logs out of it.
I want to archive all the channels I'm in, so I can search the archives offline using regular text search tools. But as far as I know there's no way to do that. (The closest I've come is copying and pasting text out of it, screen by screen, which is a very long and tedious process.)
The Discord client has rudimentary search capability but you have to be online and connected to the Discord server you want to search to use it, and there's no guarantee it'll continue to work indefinitely, and if you ever leave that server your ability to search it is gone.
Scrolling through chat history is also incredibly slow (especially if you have to scroll more than a little, as it slows down significantly when you scroll back a certain amount and it has to load the chat history from the server).
The Discord client is a resource hog too, and on my old slow laptop I dread playing a game and having Discord open at the same time as this combination will often slow down the game to the point that it's unplayable.
None of these problems exist with IRC.
With IRC you own your own data, and if you want to log and search offline it's super easy to do. IRC clients like weechat are super lightweight so don't cause any problems when running along with other apps, and scrolling back through chat history is lightning fast.
The main thing that Discord has going for it as an IRC replacement is that it can show images and videos inline in the chat, and it has a nice looking client. But having your data locked away and at the whims of a corporation and having to suffer through all the other annoyances and inconveniences of the Discord client makes it a poor replacement for IRC for me.
From a follower perspective, it is a place that usually garners like minded individuals, meaning you can often find friends/people to do stuff with. It is also a very good way to hear news about whatever you are following, a lot of server have a ton of users but little activity because most people are there for the news.
The reason why people don't just create forum websites is mostly because it costs money and secondly because going to a website is more cumbersome than opening an app that holds all these "forums".
Maybe for 24/7 gamers who have tweaked it until they found the right setup, Discord “just works”, but for me it seemed unfit for purpose.
Recording the audio in windows you can easily tell the difference between them, but not over discord. Even tried upping the bitrate but it made no noticeable difference.
Mumble on the other hand tends to be pretty reliable and comes back up in a few seconds anyway whenever it fails.
Teams on the other hand has a terrible UI for text chat, but the voice/video meetings are fine.
Maybe people are just impressed solely by UI but they are apples and oranges wrt quality.
I also admin several servers, some of which are completely unrelated to tech and non-tech people find it seamless to use.
And their noise suppression just works. I have only seen better video quality in Google Duo, and better voice quality in 4G VoLTE calls.
Only three issues I have with Discord-
1. It requires high speed internet connection. How high? People with 3 MBPS reported seamless use.
2. The clients are resource hogs. The mobile clients drop the battery too quick, and takes too much RAM in PCs. Although the latter has improved with updates.
3. I am concerned about privacy. Discord is as close-sourced as it gets.
It's a recorded (sorry) issue, they don't seem to care. My use case is simracing with a couple of button boxes in addition to those on the wheel; the outcomes of this issue can be incredibly random (and frustrating) mid race.
Also (especially applicable if you host a small community) to get high quality streams you must put a staggering amount of nitro dollars in on a monthly basis.
It's just There.
Some people just can't accept that proprietary software can be and actually is good.
Discord does not have ads either.
https://wiki.mumble.info/wiki/Audio#Echo_Cancellation
But looks like they maybe still have blockers that prevent them from enabling it by default:
https://github.com/mumble-voip/mumble/issues/4178
Normally I would attribute this to these types of libraries evolving from the UNIX mindset, which almost always lacks sane defaults. But in this case, it looks like it has more to do with difficulties around maybe Apple's secrecy and patents around their own echo cancellation implementation. Hopefully I'm wrong about that.
When I was working on a networked game around 2005, I dabbled with Speex and Vorbis. I found the Ogg maintainers to be rather hostile to my suggestions around echo cancellation. I suggested some stuff with autocorrolations to find the echo delay offset, since I had used them quite a bit as a contractor at hp. I was thinking that maybe they would let it dynamically adjust with little overhead, rather than needing a fixed delay between the microphone and speakers. I wanted this for the Mac, where there was no low-level way to access the mic and speaker hardware like on the PC, so I was concerned that there wouldn't be a fixed time offset available with Apple APIs.
At the time, I thought they were just dismissing me as a n00b, but now I understand that they were just terribly overextended. These libraries are difficult to maintain, not because of any technological reason (this stuff was well-understood by the mid 1990s), but because of hardware/driver errata, constant undermining by big established OS players caving to entertainment lobbyists like the RIAA, and the constant threat of lawsuits by patent trolls.
I would say the same thing about mesh networks or BitTorrent or Skype before Microsoft bought it or TikTok any of the other faces of the "real" internet that routes around censorship and inconveniences those who profit from the status quo.
Not that I'm advocating Teams here, far from it, I think of all the remote-meeting platforms it's one of the worst, but it's MS, so regardless how bad it is, it'll maintain it's significant market penetration.
[1] - https://support.discord.com/hc/en-us/community/posts/3600634...
It might not be transcribed, but I'd be very surprised if they didn't already save everything that everyone said (in audio format).
Some of it is probably already transcribed too. It's just that doing it for every possible accent and language is not easy without training, so they probably don't have everything transcribed, but some of it probably is and the rest is just waiting in archives for when they want to give it more attention.
In the age of everything WebRTC that only works with JavaScript and countless invasive browser "standards", Mumble is very refreshing. (Yes you can technically tunnel WebRTC through Tor by using firefox/chromium instead of Tor Browser, but by doing so you will leak your IP!)
How do you even set it up to use TOR?
It'd be nice if it just had a "Connect through TOR" checkbox in its preferences, but I have a feeling it's not even remotely so easy, is it?
I believe it's as straightforward as that, but to be honest i haven't fired up a network log to check that Mumble wasn't sending IP addresses or other "private" information over there.
So you have to set up tor too.
The average person's not going to know (or even want to know) how to do that.
It needs to be made as simple as a "connect over tor" button, and everything should be done for you, including setting up tor... anything requiring more interaction or knowledge is going to lose a lot of users.
Auto-setup may be easier if you're targeting a certain platforms, for example on Android you can keep a local copy of the F-Droid PGP key, download the latest installer, set it up, and from there enable Guardian Project repository and setup Orbot (tor for Android). On Debian, you could just run "apt install tor" (acquiring privileges on the go) from your program and then start proxying through localhost:9050. But if you're targeting many systems the problem becomes hairier.
If you have better suggestions about how to ease this process, i'm sure folks from the Tor community will be very interested! :)
Edit: Ok, that doesn't work. You're right, there should be a Tor button in Mumble.
- you slap together a low effort UI in order to test your code.
- You use this UI so much during development that it becomes "intuitive" for you.
- Early adopters are interested in functionality, so they can overlook a poor UI for a while.
- They use this poor UI so much that it becomes "intuitive" for them.
- Early adopters even help newcomers to overcome the poor UI.
- Meanwhile you develop more features and postpone the "less valuable" UI improvement.
- At some point there is so much functionality hooked to the UI that it would be a drag to change it. Also as it was quick-and-dirty stuff it would also be very difficult for others to redo the UI.
- You find a way out: pretend that it is vital to extract the library hidden in your program, so that "anyone will be able to put the UI they want on top of it so leave me alone".
- This further postpone the new UI, because potential UI makers have to wait until the library is done, and then users have to wait until the UI designer and implementer does their thing.
- It takes too long because portability issues and API design is not so easy, so you lose your UI designers one by one. So you have to make a low-effort UI to test and demo your library.
That's really hard to organise in opensource world. And when you try to introduce leadership which can do it and try to collect information, you get the issues Audacity ran into. Sure, Tentacrul can lead the UX effort and will make Audacity much better, but the community impact was pretty negative.
The only opensource project I really remember pulling off a well organised UX update is Blender. And even that was after years of people screaming "don't touch right-click-to-select, we're used to it and newbies need to learn".
Too many projects tightly couple their UI and mechanism, which always leads to problems.
[1] http://www.catb.org/~esr/writings/taoup/html/ch01s06.html#id...
[2] http://www.catb.org/~esr/writings/taoup/html/ch04s04.html
edit: found the thing where people got upset about telemetry - https://github.com/audacity/audacity/discussions/889
Quoting from a commenting on this thread:
"The only thing that can confuse people is the certificate security system... ..That is an issue of general technical illiteracy, that Mumble cannot address."
I think this sums it up. While not universal, OSS has a higher tendency to not-my-problem certain things. Often this is UX, other things too.
OTOH, OSS tends to delight at taking ownership of many other types of problems. Interoperability, multiple implementations, advanced features, user choice, etc. Stuff that proprietary software tends to be bad at.
The thing that drives facebook or Tinder to make their software addictive is the same thing that drives them to make it usable for the average person.
* Code is more interesting to the author than user experience
* High familiarity with a tool, as authors gain by working on a tool, will normalize bad UX
* Free software (specifically) often attracts a type of person that has very non-mainstream UX tastes
It takes real top-down leadership (or a one-man project) to be able to change those things.
A recent story I have been fascinated by involves the Android reader RedditSync - not OSS, but championed by a fearless Solo Dev. He released V20 after so long and so much feature feedback from dedicated users in the Play Store beta channel. Immediate community revolt from V19 users who just liked the UI the way it was and never new there was a beta program in the first place. Much subreddit infighting. Rollback. Cooling off period. Now its basically forked with X users on a 'beta'V20 and Y users on the Play Store V19. [1]
Honestly my heart goes out to the dev who solid knocks out feature requests and just didn't expect the level of cling his old UI had developed. Some people were proper angry and he was in a tough spot.
Anyhow its probably better material for a business school case study but point here is even a one man project can get caught by UI pattern bias.
IMHO its more than just lack of designers (though that's important) it's actually a balance of power thing.
Delivering really good UX requires taking a design-led approach to the whole project. Unfortunately this conflicts with one of the main reasons coders enjoy working on open source. No management, no customers and you get to work on what you want. Design it for yourself, not others.
But of course, the interface that the average coder wants is nothing like the interface that the average user needs - especially if the average coder is intimately familiar with all the features. Most coders appreciate this and try to design a 'friendly' interface but at the end of the day it's a power imbalance. In a conflict between clean design or adding more features, a team led by programmers is going to prioritize features.
When I went to design school and tried to use Inkscape I was laughed at. The day-1 was basically here's a Mac, here's Adobe, now learn to play in this sandbox (which I did til post college). FOSS tends to love Linux and BSDs support, yet most popular design tools aren't available there.
For what it's worth, I'm a design lead and make six figures. I don't think that has much to do with anything.
I do not follow this line of reasoning. There are Indian FOSS developers being paid 10 times less compared to their SV peers. Why would your salary matter? Surely, what‘s important is the amount of dispensable time you have, not your salary.
I think it is (unfortunately) a privilege of being able to afford the FOSS and privacy-focused alternatives--through money and time. A good salary gives you room for privilege. I don't think you can separate finances from the equation. Many people are just out there trying to take the easiest route to survive, and FOSS isn't as easy.
(Heck, even speaking English is a privilege many here on HN have. Many FOSS projects are only in English. I've been in Thailand for a while now and while the much of the youth demographics resents its government, almost no one knows about decentralized, private, FOSS services because it hasn't been localized and people can't afford the bill either--and as such the government has many times censored Facebook and YouTube and other centralized systems.)
Reminds me of a quote from http://theantisense.com/2018/10/26/biohacking-trash-flavored...
"I understand that “teaching a man to fish” is a thing, but that metaphor breaks down under the constraints of time and the pressures modern civilization. The ability to make tools stems from access to time. The freedom of time comes from having money. There’s a reason all those Renaissance dudes knew 7 languages, had spare time to write poetry, write essays on philosophy, and built their own laboratories. It’s because someone was doing their dishes and laundry for them. Someone else was subsidizing the overhead."
(The article make a rather different point and that quote is slightly cherrypicked, btw. It's a good article, I recommend it.)
Nothing can compete with discord where you just press a button to create a “server” and send out a link. It all just works and it costs a fortune to run.
No one has enough free time to build a brand new high quality app for mumble which works on 3 desktop OSs and 2 mobile.
"""No one has enough free time to build a brand new high quality app for mumble which works on 3 desktop OSs and 2 mobile."""
It also can't be (whatever windowing kit, E.G. QT) because that requires a big download in addition to the program, and also 'doesn't feel native' or 'doesn't look right' or 'sets off the AV scanner'.
The other escape hatch is electron; which looks "web", but is even _more_ bloated, but at least it doesn't set off the scanners. Instead it just gobbles CPU cycles, memory, and is slow and horrid.
I'm to the point that I don't care what wins, or how horrid writing for it is; I just need a widgets toolkit and bindings that can be developed for ONCE, is hosted with the OS and shared among all apps, and works on Mobile (all of them), Win, OSX, and 'nix.
I keep wondering if something like Godot could win out in GUI-land. https://medium.com/swlh/what-makes-godot-engine-great-for-ad...
So I'd wager that lots of OSS have bad UI because no commercial entity considered that the market is big enough to justify spending resources on it. This is, an external company pouring money at hiring OSS devs to work on the UI, or maybe even the project creators themselves founding a one-person business to sell services or products that require such a good UI.
An example: if Mumble had a huge user base, maybe there would possibly be some company selling high-quality, easy to use desktop applications for it. But reaching that goal is very hard, and most OSS projects never get even a fraction of the traction that would be needed for such commercial efforts.
Kudos to the Tor Project and Conversations.im for making decentralized, encrypted communications within everyone's reach.
This is probably worsened by the fact that people who care about good UX will gravitate towards software with good UX, which is often commercial.
I’d happily give them the $4 a month per employee they were requesting at the beginning for WebRTC, but they keep jacking up their prices, currently at $12, increasing by $2 every 6 months…
…and now they present login as if it is mandatory, even if we pay for it on our side, which is annoying for customers/applicants.
- Native UI: everything happens instantly
- Lightweight client
- The ability to host your own server
Even initial setup difficulties had an upside. More often than not, those who weren't capable of properly setting things up also weren't the right fit to play seriously/competitively. Such little barriers were a great initial filter for toxic wannabes with short attention span.
That's not the fault of the "heavy" protocol, but the implementation of it. As of now there's only 1 reference server implementation which is written in Python (Synapse), but https://conduit.rs/ and Dendrite are making good progress to be able to run on embedded devices.
I made https://guildbit.com years ago for gamers that want to just quickly spin up a free temporary server with their friends. I still operate it today since there's still a few users out there. I realize most are using Discord now.
I still think it would be great if someone were to create an open source Discord clone using Mumble/murmur as a backend. I'm sure it's possible, but not too sure if there's still a sizable audience willing to use Mumble over Discord these days.
It has allowed for so many cool in game features, like the ability to have DSP audio filters on channels so we can do RadioFX audio effects in game without any external applications.
[1] https://github.com/citizenfx/fivem/tree/b959f174803a972cf8c7...
Only a few "hardcore" corps are still running mumble
Thousands?
The server ran on a "free-tier" server that the hosting company gave us for free, since we were renting three of their top-tier gaming servers. It used, at peak, 5-7% cpu, and 250mb memory, if I remember correctly. Normally it was closer to 2% cpu and 100mb memory. This was with all the settings jacked up to "ultra" or the equivalent.
Audio quality, and latency were amazing. It's really disappointing to use video chat like Google Meet, or Zoom after a product like Mumble.
My servers & mumble server finally got retired, but some of the regulars started their own server, and occasionally I'll login still and a bunch of them are still hanging out. Mumble is an incredible, extremely stable product. Highly recommend.
WebRTC has seen an incredible push and if you only wanted Mumble for its open-sourceness and its use of the Opus codec for low latency, you might be better off by joining a Jitsi room with your friends and enjoying the immense effort that the web browser already brings in matter of echo cancellation and other shenanigans related to conferencing apps, plus the better UI that these kind of projects offer for end users (Mumble's bad UI is a common topic).
On the other hand there is Discord, but it plays in a different league because it is closed-source, VC-backed, and uses some advanced technology like AI-based noise cancelling.
What we tried and their pros and cons:
1. Google Meet:
* ○ Excellent noise suppression only if you are on one of their higher paid plans like Enterprise or Business Plus
* + Statistics on participant activity (how often did they enable their mic/camera, bitrate, etc.)
* - Overheating on MacBooks
* - Room will be deleted if nobody is joined for X days
* - Real full screen for screen sharing not possible
* - Proprietary, needs a paid plan
* - Forget about privacy
2.Zoom:
* - Proprietary, no privacy, security vulnerabilities
3. Teamviewer:
* - Needs a separate software (otherwise a worthy competitor to Google Meet)
4. Mumble:
* - Others in this thread have already laid out why it is not being used more: UX, setup wizard, no echo suppression, finicky push-to-talk
5. Discord:
* - No privacy
* - No encryption
* - Focused on gaming
* + Interesting feature for screen sharing: You can select if you want readability and low frame rate or high frame rate and low readability
* + Voice channels
6. We now ended up with a self-hosted k8s instance of Jitsi Meet (they host their own at [2]) which works great:
* + Open source
* + Latency and call quality is phenomenal
* + sharing screens of multiple participants works like a charm. Sharing a screen in Google Meet takes 5 seconds or so, in Jitsi Meet, it is immediately.
* ○ Echo suppression is not as good as Google Meet's
* + Unique features like synchronized Youtube player (which automatically mutes all participants)
* + End-to-end encryption
* + Recordings
[1] https://www.theverge.com/2020/10/7/21505519/slack-instagram-...
We've been really excited about the Google Meet noise cancellation, but I guess our work account either doesn't qualify or hasn't had that rolled out yet. Our gSuite account is really old, from back when they were free, so maybe it doesn't qualify. Meet works well in general though, we use it for our lunch chats.
In the past have used Jitsi, had some good and some bad experiences with it, but mostly it works well.
I just tried qTox with my wife, and had her also on the cell phone, and qTox was at least as good as the phone for latency. She was on wifi on Windows, and I was hardwired networking on Linux.
qTox uses the same codec as Mumble, "Opus".
> In the past have used Jitsi, had some good and some bad experiences with it, but mostly it works well.
Just to be clear: Did you use Jitsi or Jitsi Meet? Also, Jitsi Meet supports end-to-end encryption while almost no other solution does.
I remember Tox being very unfinished when I tried it a few years ago. Glad this has changed a bit.
Also just update your version if you are really so concerned? This doesn't affect me at all on linux and the current development snapshot for windows is using an updated openssl lib.
I don't want to run a server with unpatched known security holes so I had to scrap that idea.
"Just update your version" is also not possible because the newest static build still contains old OpenSSL.
In the end I had to get the newest murmur package from Debian Unstable. It's fine, it works for me (TM), but then why did the static build exist at all?
They should just remove the outdated static binary build if they aren't going to fix it.
At least they put a warning on their release page: https://github.com/mumble-voip/mumble/releases
But releasing a vulnerable version of Mumble is still bad. They should either fix it or not release a static binary at all.
I would like a open source solution similar to discord, I think the closest thing we currently have are Signal group calls.
Current features of Signal calls: - peer to peer no need for a central server which sells your data (Discord) or needs to be setup rented, maintained (teamspeak, mumble)
- end to end encryption
- video calls
- Screen sharing
- large group sizes for text chats (1000)
Wanted features: - adjust the volume of individual participants
- multiple voice/ text "channels" for a group (something similar to discord servers)
- usernames so that not everybody needs to know my phone number
- more participants in a single group call (currently maximum 8), I think 10-15 should be enough for most use cases, with more voice participants voice chat breaks down regardless, because of people constantly interrupting each other
- the UI for group calls is still a little confusing
Discord doesn't sell your information: https://discord.com/privacy
FTA:
> "The Company is not in the business of selling your information"
Security problems.
Barely maintained.
Some pretty major os bugs.
Most people can't get it to work first try.
I have to say, my game group uses it instead of guided or discord and it is genuinely annoying.
Mumble is a tool with its own strengths and weaknesses. It's not a social communication platform, so of course it sucks to run a community on it!
What it is is an excellent voice server. I've used it for intercom at a large live event and it was perfect. I've seen it used at a large LAN gaming event to successfully cut bandwidth usage almost in half. It also has broad support for positional audio in games - something that Discord notably completely lacks.
It's free, it's open, it's not bound to provider-operated servers, it has stellar latency and is very extendable. Please, find me another option.
Simple convenience is the killer feature. Niches will use niche software
Convenience isn't a killer feature for us - it's the actual features. Positional audio simply isn't available on Discord and depending on what you're playing, might only be available via Mumble and nothing else.
It's a tool like and other with its own use-cases. It will never be the one and only communication platform and that's a good thing!
This is the kind of stupid elitism people have here. As if I haven't hosted Ventrilo, TeamSpeak, and Mumble servers for years. Maybe these days there isn't competition (although you mentioned TeamSpeak), but even back in the day it was a tiny miracle if two different groups actually shared a platform even inside the same game different groups/clans/guilds usually had different setups and I had to juggle multiple clients just to participate compared to now when everyone is simply on Discord.
Unless you need some niche feature (like the OP's positional audio) there simply is no reason to go for out dated client/server model.
Why do you know that?
Everyone is on Discord.
What security problems?