The audio quality is great and the latency is marvelous but I can't get normies to use it with the tedious setup wizard, poor feedback cancelling from that one person who joins with the laptop speakers and mic, and the labyrinthine security model.
I've managed to convince my friends to use Mumble just as the pandemic started. The initial suggestion was to use Discord, but I luckly averted that. I think that the wizard and the defaults in general can be improved, but it is not unmanagable. The only thing that can confuse people is the certificate security system, where a friend of mine complained they couldn't connect, but it was just a "Do you want to accept this certificate" popup that was in the way. That is an issue of general technical illiteracy, that Mumble cannot address.
> The only thing that can confuse people is the certificate security system, where a friend of mine complained they couldn't connect, but it was just a "Do you want to accept this certificate" popup that was in the way.
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.
> 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
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 people do set up at least a dynamic DNS of some sort.
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™.
I have some AdBlock and Anti-Tracker stuff configured. For each single login, discord makes me solve 2 captchas, enter my password twice and click on an emailed link. Fuck them.
On mumble however, I start the client and it just works. Also, it instantly reconnects should the connection break.
This gave me a giggle, perfect sentiment IMHO. Discord is basically saying "fuck you" at least 4 times each time you login, and the only appropriate response absolutely should be "no, fuck you"
It asks me the same even without adblock or tracker-blocker, because my IP address changes often. An annoying nightmare really when you "need" something from a Discord chat quickly. I stopped using it, insist on people emailing me or chatting on Telegram.
Has anyone found a good solution for paying for captcha solutions in your browser? I'm getting tired of solving captchas, and would pay a decent amount of money to be able to click one button and have it just solve it for me.
Can take a bit of time, but needs to be non-clunky.
Do you open it on Firefox with an ad blocker etc? If yes it may come from reCaptcha and not from Discord itself (I had the same issues on websites that use reCaptcha).
However, if you use the discord app you may not need to solve a captcha.
If not for anything else, then because Discord is a service owned by somebody, Mumble isn’t. The free and open web depends on people using the Mumble model.
It's a centralized, non-free service that feeds on people's data. That's horrible enough, but i'm guessing i can't even use it with a Tor browser? Mumble works plenty fine on the Tor network.
I haven't used Discord previously so i wouldn't know, but i read they use WebRTC. If so, then it wouldn't work with Tor Browser at all, which is built without WebRTC support for privacy reason.
On top of what has already been said, I found Mumble to be a better voice com tool than Discord, sound quality is better, shortcut mapping more extensive, etc ...
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.
Seriously, I ran a community while discord was taking off and I had an infographic for how to install and connect but still spent hours every week walking people through troubleshooting.
I tried to make a custom build preloaded with the server even.
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.
I have to agree that the security model is horrendous. It's that kind of thing that when you come up with it and program it in your own code, it looks so obvious and easy to use right?
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)
A feedback cancelling solution would probably increase latency however, so it would kill the 'low latency' claim if it was on by default.
Definitely something needed to be switched on for problematic setups though, since the disadvantages would be outweighed greatly (not having feedback \ echo is a necessary feature).
It shouldn’t necessarily, you can find out how much of the speaker output ends up in the microphone input and compensate. I’m sure the models are more complicated to deal with echoes and distortion but the same approach should basically work: fit the filter offline, apply it online.
The major gripe here should be with the monkey who was using speakers in voice comms.
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.
You can make speakers and mic sound good, but you need solid software support and also a really decent mic. I have a friend who does this with a Blue Yeti mic sometimes. If you put it in cardioid mode, have the speakers not too loud, and position the mic so the speakers are behind it, you can't hear feedback on Discord. But people running mic setups this fancy are definitely the minority.
Sure, but the vast majority of this legwork is being done by a nice mic having directional capability and massive gain control. This is far from a win for discord and audio comp software. This kind of setup can be made to work on mumble too.
I tried to switch a community I once hosted to mumble. Previously teamspeak 2 was used which showed its age through high latency and low quality. TS3 was not available yet. There were so many complaints about the client that people voted to switch back to ts2. When ts3 came along they switched to that, but I already passed on the admin job by then so don't know any details about how it went. While it's not open source you can at least self host...
Oh yeah. I love Mumble as long as it's been set up by someone else. Trying to get a server running on your PC with correct security settings and then over a phone walk non-technical people through connecting to it? Nightmare, or at least it was a few years ago when I tried.
I used it (on Android the app was called Plumble) over tor, because of tor latency there were lag spikes at times, but in general, it worked quite well.
The sound quality, low latency, self-hosting, FOSS, and customization are each plus points compared to the predecessor, Ventrilo. However, something like self-hosting isn't what everyone wants. Some just want to rent a server without having maintenance. I believe that, plus the interface, is why Discord is so popular. That, plus the network effect. Which Ventrilo once had. And Altavista, for that matter.
There's dozens of hosts providing Mumble servers. People want Discord because of (in this order): network effect, access from the browser, polished UI.
The biggest contributor to latency in any VoIP solution is the underlying network latency. Even though Mumble uses modern low-latency codecs such as Opus, it is difficult to compete with Zoom/Webex/others because Zoom and Webex are not only clients, but also the backbone network. If you want to use Mumble (or any other similar service) in an international firm, you'll need to invest in a low-latency network with proper QoS to deliver an experience similar to Zoom/Webex. I guess it would be cheaper to just use Zoom/Webex in this case - we spent millions each year for our 10-20 Mbps MPLS network at one of my prior jobs (not including the cost of the engineers).
However, it might work for a smaller company.
With P2P sound there is no need for expensive slow servers serving free tier users. Scalable! But companies have a hard time letting go of control and 'value add'.
For some markets P2P is not acceptable. For example in "gaming" markets people will DoS any IP they see so hiding your IP behind a relay service like Discord is essential.
Sure but it should not be a problem for friends or teams.
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 ...
But everyone generally accesses Zoom/Webex/et al over the public internet, right? Do they implement a CDN-like series of endpoints and have some special sauce low latency network in between nodes? Genuinely curious.
discord has a higher latency and sometimes when you start a new discussion you overlap sometimes, which is akward.
but discord has one of the best noise cancellation, which makes it the favorable solution for me.
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...
