Voice Support for the Linux Client

I keep waiting till Vivox releases Linux-blob. No more dramas.

We shouldn't wait while vivox sits on its haunches. I heard that Vivox's voice had a permissive license protocol. Linden Labs is under way with a SLVoice.exe without the proprietary fat.

If only there was a link to that code, I'd be more than willing to debug and look at what is under the hood myself.

Is voice even working for the linux version right now? If so mine won't work. It shows enabled but isn't on.

No. The Linux binary for voice does not exist. Just the interface stuff in the client.

Legend has it you can make the windows binary work using Wine or Crossover, but I haven't been able to do it, even following all the instructions. I've even played around with settings and methods, but to no avail.

Adamas

The SLvoice.exe for windows will work in wine if you set up audio for wine, which in itself can be somewhat buggy.

I'm using Crossover with the audio set up, but it would be most helpful to know exactly how Wine audio is set up. It may help me debug my problem.

Adamas

For wine there is little more and selecting OSS or Alsa and the level you want, basic, standard etc, bit rate 8 or 16 or if you want an emulated driver. Best to take the standard options, there is a test button that will give a beep output to make sure it works.

Thank you. Though it isn't the answer I wanted I can stop getting mad at not having it work

According to Tofu's posts, it appears that Vivox finally supplied the missing Blob, but predictably, the Blob doesn't work, and Linden had to send it back and tell them to fix it. I still say Linden ought to dump Vivox and go with something else that's not locked up - Vivox seems to be incapable of porting their own product, or else they just don't care.

I agree. An open solution would allow LL to fix any problems and update themselves, as well as allowing motivated and experienced geek residents to help. Getting locked into Vivox's proprietary stuff only creates more problems than it solves (as we can see: months after the Windows/Mac voice rollout, there STILL isn't ANYTHING for Linux).

Adamas

There are so many errors and 'FIXME's concerning Winsock and Dsound in the usage of WINE and the Vivox Voice - client , I daresay NOBODY has this working or running under linux. Until proven otherwise ( e.g. an example of the proper Winecfg - audio - configuration ), I even conclude that Lindenlab themselves not even have a limped version running as we speak.

So .. sue me if not so ...

Update : .. well.. before i put my foot back in my mouth , i better post my results in JIRA...

yea i completely agree...

So .. my apologies to Tofu Linden. His statements and updates about the progress have been correct.

/me chuckles and goes out for a self-deserved big belgian brewski.

What Wine solution? Tofu didn't put anything like that in this thread.

Adamas

Note : The patch as mentioned in 2041 does not need to be applied manually anymore.

Okay, who here has gotten the Wine solution to work with a USB Headset? I can't get Wine to recognize the Logitech headset as an input device (does say it's an output).

The Wiine so-called "solution" doesn't work at all. The closest I've ever gotten is for the client to recognize me (and only me) in the "Active Speakers" list. I haven't gotten as far as finding out if my headset actually works with it yet, as nobody else ever shows up on the list. Even people I know for a fact are actively using voice.

Adamas

Hmmmh.. i know there has been a lotta downtime of voice yesterday.

About 'working' .. technically I had it running, active .. and as said.. i could use the service and was heard. But there is no Quality Assurance here. It was far from ideal, to my opion barely usable and for a fully working version one would still have to wait for an official client-release from LL regarding SL Voice for Linux.

To be honest .. i tried it .. tested it .. it 'worked'.. and i stopped using it. Is all.

@Drake : You got 'esd' running ? In SL the 'Esound WaveInDriver' should be shown as your input-device. So the mic of your USB device should be recognized by Esound in the audio-cfg-tab of Wine.

Ahhh, it needs the esd shim. Okay, I'll look at it.

Update please Tofu!

esd -d hw:0,0 helped to show SLVoice where to find the microphone. Otherwise it finds a dud input that just produces noise. Use your own hw:card,device params if you know where your ALSA driver input is.
In the voice config tab set the input to an Esound WaveInDriver.
I get much better quality voice output with ALSA's dmix PCM than ESD. SO.. for usable sounding voice I have the rather bizzare situation of using ESD on one side and ALSA on the other but, hey, it wheely wheely vurks

Und you hef vurked it out better ! .. Wunderbar ! XD

I still fail to see why LL is not releasing the blob they have now.

As an alpha release, with no warranty and support, mind you, we wouldn't want them to have to work on getting it to run on users' computers.

The way I see it, there shouldn't be a problem in the network side. There also shouldn't be a problem in the IPC part.
I can only see the blob having a problem with discovering/using the wide variety of available Linux audio devices.

I believe that the Linux community, as a whole, would be happier to have a blob that is hardwired to use one audio system, then no blob that tries to grok them all.

Just give us something already, so we can stop mucking about with Wine/Crossover!

Until then, a Wine step-by-step giude/HOWTO would be a GoodThing(tm)

I second that!

I've been trying (and failing) to get the Wine stuff to work, to no avail. It's like there's something in the software telling SL not to connect if there's anything running it other than Winblow$.

My offer to test what Tofu has remains open.

I'm running OpenSUSE 10.3 i386 on:
AMD 64X2
1G RAM
NVIDIA 7300GT AGP 512M
1.5/1.0 DSL

Adamas

Good observation. The connection to the SIP servers of Vivox indeed work good enough, altough relogging to them is very frequent. In the about i see this concerning voice.

Voice chat Audio coding: Polycom(R) Siren14(TM) (ITU-T Rec. G.722.1 Annex C)

This codec is most likely used but not owned by Vivox, yet part of the voice-client .. i don't know details about it. But here is the problem, i think : The decoding and mixeroutput is developed by Vivox ( and prolly for Win32 only with a port to Mac ? ) Lindenlab is not allowed to reverse-engineer it or get it's source.

Just thinking out loud here...