Whisper clear as a bell in OpenSim
Tonight’s meeting of the Hypergrid Entrepreneurs Group was an eye-opener for me and several other attendees — well, ear-opener.
We got a chance to try out the new Whisper directional voice system for OpenSim, and it was amazing.
Not amazing as the way in Freeswitch was amazing, in that you could occasionally hear someone’s voice. For free. Whoo hoo!
But smooth. Clear. Easy to use. Effective.
In fact, it works much like the Vivox-based voice does in Second Life.
I’ve spent the last few weeks in meetings in Second Life — spending several hours a week on the platform, and comparing it directly to similar meetings on OpenSim — and there were three main and significant differences between my experience in Second Life and my experience in OpenSim.
It wasn’t lag. There were plenty of times I could barely move in Second Life. It wasn’t any particular functionality, not at the level of moderate use I was engaged in — I wasn’t creating any complex vehicle physics script. No, the differences were mostly not technological at all. Mostly.
Second Life has been around for a while. It has a lot of stuff. Fancy stuff. Pretty stuff. Interactive stuff. OpenSim doesn’t have anywhere near as much stuff.
Stuff is good. Do I miss stuff when I’m in OpenSim? Yes, when it’s stuff I need. I don’t need or want most of the stuff in Second Life, so I don’t miss it. The amount of stuff that I need is actually small and manageable.
And it’s a short-lived problem. Not only is the amount of stuff available on OpenSim grids exploding day by day, but I can also hire Second Life designers to make custom stuff, or pay a little extra for them to export their stuff for me — or build stuff myself.
Even if I set up a private, behind-the-firewall OpenSim grid, I can turn on hypergrid connectivity at any time and teleport out to the other OpenSim grids, go shopping, find designers, and bring stuff back. Then I can turn hypergrid back off if I want to lock my grid back down.
Second Life has people. A million active users each month. If I wanted to talk to random people, Second Life would be the place to go.
But I don’t need to talk to all million Second Life residents. In fact, I personally only need to talk to a much smaller group of people. My colleagues. My employees. As a business user of virtual worlds, I’m not so much interested in making anonymous new friends. I’m interested in interacting with a smaller group of people whom I already know.
Similarly, a school isn’t likely to want its students to wander around Second Life’s clubs, sex shops, and nude beaches — though the students themselves may have a different opinion on the subject. Instead, a school might prefer that students and teachers interact only with one another, in the pursuit of educational goals.
If I have my own private grid, just for my company, I might start to feel isolated. What about meetings with potential partners? Potential employees? Customers? Here, again, I can turn on hypergrid — either for the entire grid or just for certain regions — and invite visitors from other grids for meetings, public events, or marketing showcases.
For me, the one big thing that Second Life has that OpenSim doesn’t is decent voice. Now, Second Life’s voice isn’t perfect. Group voice chat is tricky to enable, and doesn’t always work the way you want it to. Microphones can be hard to configure. There’s sound levels to set. And what’s with the voice morphing, anyway?
But overall, Second Life voice is smooth, easy to use, and clear — and the lip synching is pretty darn cool.
You can’t have a business meeting without voice. At the very least, the boss has to be able to hear her own voice. By “boss,” I mean me. I do like to hear my own voice. The single biggest problem I’ve had in OpenSim is that I haven’t been able to talk as much as I wanted.
My employees and colleagues haven’t been able to gain all the benefits of my wisdom and experience. Sure, I can type that stuff — but I talk much faster than I can type. Much, much faster.
Yes, typing means that I can use Google Translate. This came in handy on a recent meeting on FrancoGrid, when I realized that my high school French class was 20 years ago and all I had left was “voulez vous couchez avec moi ce soir” from Moulin Rouge. Which, I believe, means, “Are we meeting the quarterly budget numbers?”
With Google Translate, I can cut-and-paste the transcripts and see what people are saying — or use one of the many available tools that will handle the translation instantaneously — as long as everyone spells everything right.
And there are members of my staff who are not native English speakers and would appreciate being able to read things instead of hearing them.
But the availability of voice doesn’t prevent us from typing. In fact, in the FrancoGrid, the meeting started out in spoken French and switched to typing because not everyone present could understand that was being said.
So I’ve been eagerly looking for voice solutions for OpenSim.
For about a year now, we’ve had FreeSwitch. It wasn’t directional — if you were wearing stereo headphones, you wouldn’t be able to tell where a voice was coming from. And it didn’t get louder the closer you were to the speaker. FreeSwitch didn’t have voice synchronization. Or those cool green lines coming out of the speakers’ head, so you could see who was talking.
But those are all minor quabbles. The big one was that FreeSwitch was just bad. Sound quality cut in and out. The voice would come on and stop suddenly. It was too difficult to use for real business meetings. It was okay for casual get-togethers and non-mission-critical events. But for real work? Not so much.
The great big hope of OpenSim has long been Vivox. That’s what Second Life is using. Vivox provides high-quality voice to big-budget multiplayer video games.
Unfortunately, OpenSim is small potatoes. All the public OpenSim grids put together have about a tenth of the users of Second Life. (The Hippo viewer, the most popular in OpenSim, has about 80,000 users.) And the public grids have about a third of Second Life’s land area.
This is bad enough. What’s worse is that all these users and regions are scattered around a hundred public grids, and who knows how many thousands of small, private closed ones. Vivox would have to sell the service to each grid and hosting provider separately. The support calls alone would be a nightmare.
A few months back, there was a rumor that Vivox was considering letting OpenSim providers have Vivox, at the low, low, low price of $50 per region.
OpenSim hosting for a moderate-use region runs at around $25 a region, so adding voice would triple the cost of a region.
I would be willing to pay that price for high-end voice — and it would still be cheaper than Second Life. But I couldn’t. Vivox voice never materialized on OpenSim. Maybe in the future, OpenSim would grow to the point that it was a viable market for Vivox.
I was starting to try to think of work-arounds. For example, I want to hear myself speak, but I’m not so anxious to hear other people talk. So if I was speaking, and everyone else was typing — well, I could just talk for a whole meeting without interruption! I could do this by piping in a webcam feed over the new media-on-a-prim support that OpenSim developers rolled out in their latest release of OpenSim.
Now it looks as though I’m not going to have to go to that extreme.
The Whisper voice service is a free, open source product that acts, looks and sounds like Vivox. Directional voice. Little green lines over the heads of active speakers. Even lip synching, if you’re using a new viewer that supports it. And it’s crystal-clear. No cutting in and out.
Snoopy Pfeiffer of Dreamland Metaverse has it up and running on the “Mumble Sandbox” region on OSGrid, if anyone wants to see it for themselves.
Whisper is only available for the Windows platform right now — Mac and Linux will come later. For it to work with the viewer, the Whisper software needs to be installed in to the same folder as your viewer software — such as Hippo or Imprudence. Instructions are here.
Eventually, Whisper may come bundled in with Hippo or Imprudence so that users don’t have to download it separately.
A second component, called Murmur, needs to be installed as a module in each region’s server software. This is something that would normally be handled by the hosting company, your company’s tech guy — or by a talented at-home tinkerer.
According to Volker Gaessler, founder of vComm Solutions of Switzerland, the server-side module will be released as open source in the “next weeks.”
At that point, Second Life will lose its last remaining advantage — at least, as far as enterprise and education users are concerned.
Now, voice does add additional load on the server running it. This may or may not impact the performance of your region. With today’s test, we had about ten people in the region, most of us either talking or listening, and things worked fine.
It would be nice to see some larger-scale performance tests to see how many people the platform can really handle.
And there were a few glitches. For example, since the software was just released, it’s not packaged in with Hippo or Imprudence yet, and has to be downloaded separately. Also, it doesn’t immediately disconnect people when we leave a region — after we teleported out, we could continue to hear people we left behind, and they could hear us. This is likely to get fixed before the software is officially released.
Overall, though, this was a major step forward in usability.
Here is my prediction for enterprise uses of OpenSim: With media-on-a-prim, Whisper voice, and the more secure Hypergrid 1.5, OpenSim is now good enough to use for light-to-moderate enterprise applications. This means classes and training sessions, internal meetings, and small group presentations.