Voice/Linux Sound Settings

Sound Settings for Voice on the official Linux Client

Do you use voice on the Linux Viewer? If you do, please add the sound settings that work best for you.

We will tabulate the settings; for now, please add

• the linux distro and version you are using
• relevant hardware specifications > CPU type, sound card type etc.
• headphone or microphone type (USB, via soundcard etc.)
• the sound system architecture you are using: ALSA, ESD, or OSS
• the specific sound card driver your sound system is using (for example: snd_hda_intel)

Thank you!

Reported Settings

Ubuntu

• Ubuntu 7.10
• listening: ok, some glitches while zooming around
• speaking: not tested
  • motherboard: Asus P5B SE
  • CPU: Intel Dual Core
  • sound card: 82801H (integrated)
• speakers via sound card / no microphone
• ALSA 1.0.14-1ubuntu2
• snd_hda_intel

Ubuntu

• Ubuntu 7.10
• listening/speaking: NILL (white dot but can not hear or speak)
• Alsa is selected in devices under voice settings.
  • motherboard: Asus K8vSE Deluxe
  • CPU:AMD athlon 3000+
  • sound card: (on board chip) VT8233/A/8235/8237 AC97 Audio Controller
• speakers and headset via sound port, no USB devices
• ALSA 1.0.14-1ubuntu2
• Esound version 0.2.38
• OSS

Debian

• Debian unstable --Tx Oh 15:10, 23 February 2008 (PST)
• listening: ok
• speaking: ok
  • notebook: lenovo N200 3000
  • CPU: Intel Core2 Duo
  • sound card: 82801H (ICH8 Family) HD Audio Controller (rev 03) / snd-hda-intel
  • kernel: 2.6.24 (http://kernel.org/)
• speakers and microphone via sound card
• ALSA via kernel
• pulseaudio version 0.9.9-1
• with the following work arround in the Secondlife directory:

$ cp SLVoice SLVoice.bin
$ cat > SLVoice << EOF
#!/bin/sh

padsp ./SLVoice.bin \$*
EOF

Debian Etch

• Debian Etch

Voice (listening and speaking) works ok, though the music stream stalls so every now and then.

Debian Etch with a custom kernel from kernel.org; 2.6.24.

AMD Sempron 1.6 ghz (overclocked to 2.0 ghz)

nVidia Corporation nForce2 AC97 Audio Controler (MCP) (rev a1)

.config: Disable OSS. Enable ALSA. Both at once caused troubles on my system with Secondlife Before starting Secondlife, the OSS emulation modules snd_pcm_oss and snd_mixer_oss had to be disabled (rmmod as root) as they were interfering with nVidia (AL lib: alsa.c:342: Could not open playback device 'default': Device or resource busy). After that, it worked ok. This is fiddly, and you might wait to re-add the modules (modprobe as root) after you're finished with SL, as other applications might depend on them.

Ubuntu

(hoisted from the sl dev mailing list)

• Ubuntu 7.10
• listening: ok
• speaking: ok
  • Running Ubuntu 7.10 and my experience has been that sound is finicky and voice doubly so. ESD does seem to be the way to go. Here's a breakdown of what I've tried:
  • Kernel runs ALSA modules including OSS emulation
  • RC 1.19.1 (1) auto-chooses OSS according to the console and has functional but degraded sound out of the box (music is to choppy to listen, but UI and general sounds can be heard). Voice doesn't work, console shows "AL lib: alsa.c:342: Could not open playback device 'default': Device or resource busy"
  • README-linux-voice.txt says that voice uses ALSA. Okay, lets ditch OSS emulation.
  • export LL_BAD_OSS=x gets me "ALSA audio output initialized OKAY" Cool
  • Actually, it's not cool, in ALSA mode all of the normal sound in SL is broken (no audio at all) but lo and behold, voice works!
  • I see Alan's message [on sl-dev, ed.] and figure okay lets try esd. "ESD audio output initialized OKAY" Yay.
  • Sound effects work but are delayed (esd buffer settings maybe?). Double yay, music is listenable! It all sounds great. Triple yay, Voice works! Hopefully this will help other linux users become fully audio/voice capable in SL.

Fedora 6

• Fedora 6
• listening: Usually clean
• speaking: not tested
  • motherboard: Asus
  • CPU: AMD X2 6000+
  • sound card: nVidia MCP55 (Intel_HDA) (integrated)
• speakers via sound card / no microphone
• ALSA 1.0.14
• snd_hda_intel
  • Works beautifully on SL Client 1.19.0.5, constantly reconnects on SL Client 1.19.1.4, SL Client 1.20.0 RC

Fedora 8

• Fedora 8
• listening: Dead in the water
• speaking: not tested
  • motherboard: Asus
  • CPU: AMD X2 6000+
  • sound card: nVidia MCP55 (Intel_HDA) (integrated)
• speakers via sound card / no microphone
• ALSA 1.0.15-1.fc8
• snd_hda_intel
  • Nothing on SL Client 1.19.0.5, constantly reconnects on SL Client 1.19.1.4, SL Client 1.20.0 RC

Technical Background

Voice in Second Life uses Vivox technology that is based on RTP (using the oRTP library), SIP (using the amsip library from Antisip), OpenAL, TinyXPath, OpenSSL, and libcurl for the transmission of voice data. The positional 3D audio is achieved with technology from DiamondWare. These technologies are contained in external daemon software that is started and stopped by the Second Life client.

You can find the official ViVox documentation on the JIRA.

The SecondLife viewer handles configuration, control, and display functions, but the voice streams (from the microphone and from the Vivox voice server) do not enter the viewer. Unfortunately the source code for the external SLVoice daemon is not available due to contractual obligations between Linden Lab and Vivox.