Observation of voice functionality

I had quite a few problems getting voice to work initially.. Background details as follows:

Computer has 3 sound devices:
1) On-board sound chip (SoundMAX) for as the windows default sound device
2) Marian Marc-8 MIDI for recording work
3) Trust USB (C-Media sound chip) headset for Skype & SL voice.

At first I couldn't get voice to work even when I selected the USB headset using the voice configuration utility in the SL viewer preferences. It would go through all the motions, I'd have the little while spot over my head and be able to hear my voice in my headset but not have any voice communication with any of the other avatars.

It wasn't until I selected my headset as the default sound device in the Windows 'Sounds and Audio Devices' Control Panel applet that it all picked up and worked properly. The downside was that all the SL background sound & Windows system sounds were also routed through my headset. This was a repeatable phenomenon on my laptop (Dell Inspiron 6400) in that voice would not function correctly unless the headset was selected as the default sound device in the Windows 'Sounds and Audio Devices' Control Panel applet. In all cases the drivers are bang up-to-date [caveat: The Trust USB headset was not shipped with drivers but uses default windows drivers]. It does seem like a routing problem. Curiously Skype works just fine with the built-in sound chip as the default sound device and the headset as a secondary or tertiary sound device.

Has anyone else come across this phenomenon? Thoughts, suggestions, etc., welcomed.

Does the voice application give you a selection of input/output devices to choose from? If not it's probably hard-coded to use the default windows driver. This is not hard to understand since LL obviously rushed this out as if at gunpoint and knew in advance that they wouldn't have time to do proper driver compatibility testing. It stands to reason that by using whatever driver the system is using, there is already a translation bridge between the system and that driver, so they're leaving that translation work to Windows itself.

I dunno. But it sounds real good doesn't it? LOL

I’m not sure what the problem is. How would you prefer to use your headset? Are you wanting the ambient sounds to come out of external speakers? Not hear them at all?

Kascha, the voice configuration applet in the SL viewer configuration screen has a drop-down pick-list that looks as if it allows you to select and specify a device for voice.

DaQbet, I suppose I would rather have the SL background sounds coming thru the speakers attached to my computer rather than the headset (which I would prefer to use for voice only). But really having voice and background sounds separate is no biggie. What surprised me most was that the SL voice software would only kick in if the headset was selected as the default sound device in both the SL viewer voice configuration applet *and* the Windows 'Sounds and Audio Devices' Control Panel applet.

Skype, fortunately, does not have this problem. I use my computers primarily for recording / audio composition and normally have items such as USB headsets, on-board sound chips either disabled or else set as the secondary, tertiary or otherwise sound devices. So basicaly. It means that I'll have to change my computer's confiuration each time before I fire up my SL viewer and then put it back the way it was afterwards. Again, no biggie, but it just seems like so much unneccessary palaver. I can actually make a skype phone call through my computer while recording a synthesiser piece and it works fine. The audio streams for skype and my recording software are kept separate and do not overlap or merge. No trace of the music appears in the skype phone call and similarly no trace of the skype call turns up in the recorded music.

I believe the special 3D component is what really makes the big difference between a skyp type applications and SL voice. Skype is flat and separate from the SL environment so not so invasive to your Windows/audio configuration. But integrating voice into SL, Vivox had to provide a total emersion type experience with the voice component. I ran into the same issues when I first started using voice. My system is controlled by the RealTek audio controller and was a bugger to figure out, (actually it’s quite easy I just made it too complicated). This is also a problem for voice morphing software. It can take some fussing around to get it to work, but that’s was prompted my first comment. Many people are having issues getting voice to work and your post was a very thorough explanation of how to configure voice. I was thinking “taa daa” and you were thinking “booo” LOL

All the mixing is done on the server, not the client, so the SL voice audio is just as "flat" as Skype's is.

AFAIK it's SLvoice.exe that actually outputs the audio as well, so there really is no difference between a third-party proggie, or SL's voice implementation.

What makes you think the mixing is done on the server? And what difference would that make when the server has no ability to talk to the client hardware to shape, render, and distribute the signal? You know - the grunt work that sucks CPU time and RAM.

SLvoice.exe runs client side, as it would have to to have the authority to give direct commands to the system drivers. It is the client that has to render the sound and calculate how to handle distribution of that sound to one channel or the other based on a particular system's low level driver, not the server. The server sends 3D coordinates to the client app which the client app uses to place the sound. Any way it goes, 3D sound rendering is light years different than transmitting mono or stereo 2D sound.

Skype does not use positional audio. You couldn't be more misled in the comparison of Skype and SLVoice from the standpoint of resource usage.

Try creating a simple 2 minute animation in something like 3D Studio Max or Bryce, rendering it, and then converting it to a .wmv. Now do the same exact thing with background music embedded. Come back and tell us how much longer it took the second time and how much larger the file size was. Then carefully consider the fact that the sound you added did not have rendered depth and imagine how much time would have been added if it did.

How many of the devices you listed are capable of 3D audio? Leave out the midi device right off the bat, but is the SoundMAX chip capable of 3D rendered sound? I know C-Media makes 3D sound capable chips and it might be that this is the only driver/chip setup on the system that can handle positional audio.

Among other things, this quote: "we have optimized it so that all voices are condensed into a single mixed voice channel, customized for each client.".

It happens the other way around. The client sends positional and/or camera data to a server at Vivox which renders the audio different for each and every listener and then just sends the mixed audio stream back to the client who only has to decode the codec and play, which is no different than listening to an MP3 stream (from the client's point of view).

Excuse me? Excuse me! Ma'am! I believe I said I prefer my crow WELL DONE! This is clearly RAW!

But this still leaves a resource problem on less than optimal clients (albeit a much less devastating one than would be in effect if the rendering was completely done by the client). From what you are saying, the voice signal appears to be coming in in similar fashion to the parcel sound that SL sends, is that right? The performance of that too seems to vary widely depending on processor load and has been reported to be inconsistent - going in and out with some regularity.

The plot thickens

I have always had a *really* poor framerate and all, and since adding Voice my experience hasn't gotten any worse. Just my personal experience.