Open Source Meeting/2007-09-20

From Second Life Wiki
Jump to: navigation, search

Open Source Meeting for September 20, 2007. This meeting was held in using the new Voice feature.

An audio recording of this meeting is available.

Transcript is lightly edited to make it less confusing and to clean up a couple real-time mistakes (though many remain). Rob Linden was transcribing in real time for Joe Linden, so Rob's name has been replaced with Joe's in most cases.

[14:10] Rob Linden introduces Joe
[14:10] Joe Linden opens with stats
[14:10] Wyn Galbraith: LOL nice choice of words.
[14:10] otakup0pe Neumann: wow dsl sucks :s
[14:10] Dale Glass resumes vmware VM
[14:10] otakup0pe Neumann is rocking > 100% packet loss
[14:10] Joe Linden: when we went through First Look, we slowly indroduced voice over a six week period
[14:11] Joe Linden: from August 6 on, we left it up to the voice owners to enable/disable voice
[14:11] Liana Linden: Try this different stream if you don't have Voice http://205.188.215.227:8080
[14:11] Joe Linden: there is a significant portion that have opted out (not a surprise), but we;'re using it today
[14:12] Joe Linden: some numbers: in terms of estate count: 5284 estates
[14:12] Joe Linden: 1185 disabled
[14:12] Wyn Galbraith: Joe is breaking up a little. Thanks for posting those numbers Rob.
[14:12] Wyn Galbraith: 1567
[14:12] Joe Linden: regions: 14500 voice enabled 1567 disabled
[14:13] Joe Linden: since 450,000 voice enabled accounts since Aug 2
[14:13] Joe Linden: 23000 concurrent voice
[14:13] Joe Linden: 16800
[14:13] Joe Linden: 72000 unique logins per day using voice
[14:13] Tao Takashi: well I log in but might not use it
[14:14] Joe Linden: 40-45% are active voice
[14:14] Wyn Galbraith: I was just going to bring that up.
[14:14] Wyn Galbraith often logs in with voice activated but has computer muted and headset off.
[14:14] Joe Linden: 65% of the voice channels have 5 or more people in them
[14:14] Tao Takashi: no monologs ;-)
[14:14] Joe Linden: 685,000,000 minutes of voice have been used since launch
[14:14] Barney Boomslang: yeah, the install pushes ppl to activate voice - so they will do, but not actively use ncessarily
[14:15] Joe Linden: now to open it up for questions
[14:15] Rob Linden: (btw, I'm transcribing for Joe unless otherwise stated)
[14:15] Multi Gadget v1.52.0 by Timeless Prototype
[14:15] OWENIMATIONS DUO DANCER: OWENIMATIONS DUO DANCER v3.4 -- type /99help for instructions
[14:15] Rob Linden: (I missed the question)
[14:16] Wyn Galbraith: I asked how the numbers are gathered basiclly because I often log on with it activated but not using it.
[14:16] Wyn Galbraith: So one has to be in active conversation to be gathered.
[14:16] Dale Glass: Liana, authentication failed on the stream
[14:16] Joe Linden: minutes are counted by person minutes (3 people taliking for one minute: 3 minutes)
[14:16] Tao Takashi: yes
[14:16] Tao Takashi: (I think ;-) )
[14:16] Wyn Galbraith: Mostly I don't even listen. Computer might be muted.
[14:17] Joe Linden: listening but not talking is still active by this tracking
[14:17] Wyn Galbraith: Ok I got it.
[14:17] Rob Linden: general misperception possible about voice use
[14:17] Rob Linden: jody huet: is there any ability to use voice outside of Second Life?
[14:18] Joe Linden: yes, we have the ability to invite those outside in
[14:18] Wyn Galbraith: Wow.
[14:18] Joe Linden: we have a feature in beta to allow PSTN bridging of conferencing
[14:18] Joe Linden: this is standard SIP so that standards complaint systems can interoperate
[14:19] Joe Linden: there are three main modes. 1. proximal voice
[14:19] Joe Linden: that requires physical proximity to hear
[14:19] Joe Linden: 2. Group voice chat
[14:19] Joe Linden: not 3d, but anyone can be anywhere to speak
[14:20] Joe Linden: that can interoperate with those outside the SL grid
[14:20] Joe Linden: we're working on a thin client to allow use without the full client
[14:21] Joe Linden: 3. third mode: point to point talk. server is used for handshake, but after that is purely peer to peer
[14:21] Wyn Galbraith: Does that include Skype? Could someone call into SL using that program?
[14:21] Joe Linden: not today
[14:21] Wyn Galbraith: Tomorrow LOL jk
[14:21] Joe Linden: it's possible for something like that to be built, but not by default
[14:22] Joe Linden: we're using a wideband codec G.722.1
[14:22] Joe Linden: most voip systems use non-wideband codecs, so in order to interoperate, we need to transcode
[14:22] Squirrel Wood: I assume you are aware that the voice quality actually allows for people to stream music through the system and do live DJing that way?
[14:22] Wyn Galbraith: Thanks.
[14:23] Joe Linden: transcoding is very expensive computationally, so gotta be careful
[14:23] Rob Linden: Mm Alder: what about open sourcing the client?
[14:23] Joe Linden: the decison to add voice was made before we open sourced
[14:23] Joe Linden: so we're playing a little bit of catchup in making our voice and open soruce strategy aligned
[14:24] Joe Linden: all that said: we've made it clear to Vivox we're committed to open sourcing all parts of the client
[14:24] Joe Linden: you're seeing this all play out maybe a little more publicly than we'd like
[14:25] Barney Boomslang: ah. good to hear. that language was scary :)
[14:25] Tao Takashi: seen that
[14:25] Joe Linden: with respect to the recent documentation, Vivox has loosened up on the licensing text
[14:25] Tao Takashi: https://jira.secondlife.com/browse/VWR-2029
[14:25] Rob Linden: (thanks Tao)
[14:26] Joe Linden: to answer the question: we're committed to open source in all respects, adn we're working with our partner Vivox to address that
[14:26] Jason Swain: Will you be implimenting an AGC (Automatic Gain Control) on the client to autoadjust the volume of other people talking?
[14:26] Joe Linden: it's not goign to happen overnight
[14:26] Dale Glass: new terms seem to be a huge improvement :-)
[14:26] Tao Takashi: plugins! :)
[14:27] Joe Linden: our relationship with Vivox is non-exclusive, so there are lots of possibilities
[14:27] Joe Linden: Vivox is a willing partner as well
[14:27] Joe Linden: I'm happy to share a little more of our roadmap
[14:28] Rob Linden: (agreement for a couple of people)
[14:28] Joe Linden: our implementation is flexible in a nubmer of ways
[14:28] Tao Takashi: cool
[14:28] Joe Linden: voicemail is aviable on the beta grid
[14:28] Joe Linden: on the beta grid, if you make a peer to peer call, and not connect, it'll make voicemail
[14:28] Tao Takashi: me needs to more headsets, I have too many computers everywhere ;-)
[14:29] Joe Linden: we're also experimenting with voice to text
[14:29] Tao Takashi: would be cool to post that to a blog :)
[14:29] Tao Takashi: via that way to a blog I mean
[14:29] Joe Linden: we're working with a company called Spinvox to take short messages (voicemails) to text
[14:30] Joe Linden: the ability to match voice to avatar design is important as well
[14:30] Barney Boomslang: well, for a start it would be good to have just a simple VST interface in the client
[14:30] Joe Linden: we're experimenting with a voice font technology that's much more sophisticated
[14:30] Joe Linden: two reasons not to wait for this before launching
[14:30] Barney Boomslang: that way we could hook up VST processors into the mic ... (and get some voice morphers working there)
[14:30] Joe Linden: 1. voice fonts make it harder to understand
[14:30] Tao Takashi: I just need that hollywood star setting
[14:30] Squirrel Wood: using morphvox here which works exceptionably well for me.
[14:31] Barney Boomslang needs grating stone sounds ...
[14:31] Joe Linden: 2. number of parameters to tweak are quite numerous
[14:31] Joe Linden: it is getting better, though
[14:31] Baba Yamamoto: you can rest soon Rob ! ;)
[14:31] Joe Linden: event based translation of speech based options is on the map
[14:32] Joe Linden: people can opt in to recording, and a transcription can be mailed back to participants
[14:32] Tao Takashi: very cool stuff
[14:32] Joe Linden: we're investigating the ability to extract individual tracks and custom mix
[14:33] Joe Linden: so that you can do a custom postprocess step
[14:33] Tao Takashi: creating a multi track recording for later post production might be nice :)
[14:33] Wyn Galbraith: Tivo for voice.
[14:33] Barney Boomslang: well, tivo like feature for audio would rock - especially for non-natives, to scroll back and rehear parts
[14:33] Joe Linden: we're also looking at the abiltiy to have a short continuous loop to record so that people can walk away
[14:34] Joe Linden: SMS messaging out of Second Life is on the roadmap
[14:34] Baba Yamamoto: Non spatial sim/estate voice outside of groups and conference?
[14:34] Rob Linden: Mm Alder: how is client recording working?
[14:35] Joe Linden: using an outside app to record today
[14:35] Wyn Galbraith has wanted Tivo for radio iRL.
[14:35] Joe Linden: looking at giving a built in option to record a loop
[14:35] Joe Linden: client side recording use cases are quite different
[14:35] Joe Linden: both client and server are important
[14:36] Rob Linden: (missed the question)
[14:36] Squirrel Wood: I'd like to see an option "Allow others to record my voice" or mayhaps.. "xx would like to record your voice. Yes/no" dialog...
[14:36] Joe Linden: we think it makes more sense to record at the server
[14:37] Joe Linden: everyone would need to opt in tot he event, and then everyone would be offered a recording, and it would have metadata
[14:37] Wyn Galbraith: LOL
[14:37] Tao Takashi: what about different languages when transcribing?
[14:37] Joe Linden: transcription available as well for long events. not for high stakes legal stuff
[14:38] Jason Swain: Will you be implimenting an AGC (Automatic Gain Control) on the client to autoadjust the volume of other people talking?
[14:38] Rob Linden: Mm Alder: for machinma?
[14:38] Joe Linden: sure
[14:38] Joe Linden: re: Jason. that's there today
[14:38] Joe Linden: there's an energy detect floor, which detects the minimum threshold
[14:39] Wyn Galbraith: That would be good.
[14:39] Joe Linden: we also give you the abiltiy to automatically adjust the gain for individual speakers
[14:39] Joe Linden: you can grab the name of the person in the communicate window to control the volume for each individual
[14:40] Rob Linden: Squirrel: the problem is the client forgets those settings
[14:40] Jason Swain: Yes thank you squirrel
[14:40] Joe Linden: correct, it's not persistent
[14:40] Joe Linden: we're trying to make it unobtrusive
[14:40] Wyn Galbraith: That's why I added that to the jira, gang the volume when needed so that you can raise everyone, or lower, and then do an individual volume.
[14:40] Joe Linden: the mute states for individuals does persist, but not the volume settings
[14:41] Joe Linden: but hey, these are issues you should have the access to make these features yourselves
[14:41] Rob Linden: Jody Huet: are you looking at features for businesses
[14:42] Joe Linden: yes, we're looking at ways of allowing moderator capabilities before we consider our initial implemenation "done"
[14:43] Joe Linden: we're looking at ways so that 300 person groups will be able have granular microphone control so that there can be a speaker and moderator backchannel
[14:43] Barney Boomslang: good to hear. because with the current state, voice is no fun for events with many ppl :)
[14:43] Tao Takashi: Hey Zero!
[14:43] Barney Boomslang: (where many is more than maybe 5 LOL)
[14:43] Joe Linden: we think that'll help businesses have large voice conferences
[14:43] Tao Takashi: your slideshow presenter lost all the textures when I tried to use it
[14:43] Wyn Galbraith: Absolutely.
[14:43] Dale Glass: how's the Linux version going?
[14:44] Joe Linden: for large sims, you reall need more control
[14:44] Wyn Galbraith waves at Gigs and Zero.
[14:45] Tofu Linden: Hi!
[14:45] Dale Glass: hi :-)
[14:45] Joe Linden: re: linux. Tofu is here
[14:45] Joe Linden: Tofu has the SDK, and is working on it
[14:45] Tofu Linden nods.
[14:45] Tao Takashi: good news :)
[14:46] Tofu Linden: Sounds about right.
[14:46] Joe Linden: it's really Tofu's call when that will roll out
[14:46] Joe Linden: measured in weeks not months most likely
[14:46] Baba Yamamoto: 30 people isn't large enough for an ad-hoc event. For large events like Linden Town Halls have been not everyone can join an event group. How would you handle large events like that?
[14:46] Tofu Linden: I haven't evaluated the latest drop - only just got it - but the previous one was not far from good enough for an initial public release.
[14:47] Dale Glass: drop? So Vivox is working on it as well?
[14:47] Joe Linden: i.e. does it make sense to join large groups for large events?
[14:47] Baba Yamamoto: 25 groups! no more !
[14:47] Barney Boomslang: uh. "join a group". could we please get more groups before you come up with that idea? ;)
[14:47] Wyn Galbraith: There's a problem with that. Can't drop anymore would have to send an alt.
[14:47] Joe Linden: we realize the group limit
[14:47] Joe Linden: so we're thinking about transitory groups
[14:47] Tao Takashi: there might also be need for some announce only groups
[14:48] Joe Linden: for this type of thing
[14:48] Tofu Linden: Dale: Vivox develops the SDK, drops it to us, we integrate/test it with the client.
[14:48] Baba Yamamoto: That sounds fairly reasonable
[14:48] Mm Alder: When you put in voice fonts, will it be in a form that's easily swapped with a plug-in?
[14:48] Barney Boomslang: well, I guess most ppl do it like me: keep one or two groups free for projects. but they are changed round regularily - so even if I am only on 23 groups, I am maxed out - if I have additional projects
[14:48] Joe Linden: we lookeda t the stats, and a surprisingly small number of people are up against teh 25 limit
[14:48] Wyn Galbraith: :P
[14:49] Jason Swain nods at Barney Boomslang, "Yes I have the same take on the groups"
[14:49] Joe Linden: re: Mm: you'll be able put your own fonts in the viewer
[14:49] Tao Takashi: Q: What happened to the fee discussion which was going on in the beta. Has this been dropped?
[14:49] Barney Boomslang: maybe again my question from earlier: how about some standard interface, too?
[14:49] Barney Boomslang: something like VST in the voice chain?
[14:49] Tao Takashi: paying for voice etc.
[14:49] Joe Linden: re: Tao
[14:50] Joe Linden: our thinking on fees, as we've implemented it, we want wide penetration for this feature
[14:50] Joe Linden: we don't want a cost constraint
[14:50] Wyn Galbraith: People would stop using voice if they had to pay, I think.
[14:50] Tao Takashi: true, Wyn
[14:50] Joe Linden: the actual cost of providing voice is something we're subsidizing
[14:51] Wyn Galbraith: Ok. That might be different. :)
[14:51] Gigs Taggart: does this mean the plan is to not go through with extra fees for people paying $195/month in December?
[14:51] Tao Takashi: seems so
[14:51] Wyn Galbraith cheers!
[14:51] Tao Takashi: sounds good :)
[14:51] Joe Linden: we're looking at a $10/month fee for group voice, but otherwise no fees for the forseeable future
[14:52] Wyn Galbraith: 10$US per month.
[14:52] Tao Takashi: what about paying for chat instead? ;-)
[14:52] Wyn Galbraith: LOL Tao.
[14:53] Joe Linden: re: GIgs. we have not decided there's a price increase for this. we're not looking at changing any of the grandfathering of pricing
[14:53] Tao Takashi: heh :)
[14:53] Wyn Galbraith: Per character charge.
[14:53] Joe Linden: re: Tao. we've joked about charging for not enabling voice ;-)
[14:53] Tao Takashi: I forgot my question from earlier ;-)
[14:54] Rob Linden: Mm Alder: any chance of getting access to the audio streams before open sourcing the cvoice client
[14:54] Joe Linden: the client doesn't have access to the voice stream
[14:54] Joe Linden: we don't need to bring the stream into the client per se
[14:55] Tao Takashi: oh yes, I had a question about different languages for transcribing. I guess only english will be supported for now
[14:55] Gigs Taggart: wait, joe is talking THROUGH rob? :P
[14:55] Barney Boomslang: bummer. that means VST or AU plugins would have to be supported by slvoice.exe, not by the client ...
[14:55] Joe Linden: we're documenting how to interact with slvoice
[14:55] Gigs Taggart plays the twilight zone theme
[14:55] Rob Linden: Gigs yes I'm transcribing
[14:55] Gigs Taggart: k :P
[14:55] Baba Yamamoto: those greedy germans charging for software ;0 *shakes fist at Tao*
[14:55] Tao Takashi: I should directly have tried a german product instead of Hijack Pro ;-)
[14:55] Tao Takashi: Hijack Pro does charge, too
[14:55] Rob Linden: (missed question)
[14:55] Jody Huet: Question: Will specifications be released for other grids developed with the open-sourced SL server to connect with voice in SL?
[14:55] Tao Takashi: and it does not even work ;-)
[14:56] Tao Takashi: first we need to have other grids I guess :)
[14:56] Wyn Galbraith: Well, Joe sounds pretty clear except for occasional breakup.
[14:56] Gigs Taggart: not in linux, Wyn :P
[14:57] Joe Linden: re Jody. that's a hypothetical for me. we haven't thought through that in a way I can answer that in a meaningful way
[14:57] Joe Linden: Vivox is working with us to meet our needs to fulfill our commitment to the open source community
[14:57] Tao Takashi: so what about different languages for transcribing?
[14:57] Joe Linden: we're working with them on open sorucing the client piece
[14:57] Tao Takashi: I mean if the conversation is in german
[14:58] Tao Takashi: I meant more the other way round ;-)
[14:58] Joe Linden: re: Tao. yes, we'd like to offer differnt language translation
[14:58] Baba Yamamoto: Muahahaha
[14:58] Liana Linden: transcription, not translation
[14:59] Joe Linden: re: Tao 2. well over 60% of our user base is outside of the U.S., so any service we provide will be in multiple languages
[14:59] Wyn Galbraith: LOL
[14:59] Barney Boomslang: / yeah, long live google translate ... ewww ...
[14:59] Rob Linden: Squirrel: we're going to be looking forward to some very funny translations given current quality
[14:59] Tao Takashi: will it automatically create summaries? :)
[14:59] Barney Boomslang: / actgually those automatic translations are useless for anything if it is german for example translated into anything else than german ;)
[15:00] Joe Linden: this isn't for high stakes stuff, obviously
[15:00] Wyn Galbraith: Everyzing?
[15:00] Soft Linden: / German's fun. Linden = Lime Trees
[15:00] Wyn Galbraith: LOL
[15:00] Joe Linden: everyzing.com has some interesting transcription technology
[15:00] Tao Takashi: the major of the city I live in is J?rgen Linden ;-)
[15:00] Joe Linden: wrapping up
[15:00] Dale Glass: wasn't "Linden" a kind of tree anyway?
[15:00] Wyn Galbraith: Thanks for the meeting. Joe, it was great.
[15:00] Squirrel Wood: on the blog ?
[15:00] Tao Takashi: great, thanks :)
[15:00] Joe Linden: we'll post a recorded mp3 and post this transcription
[15:00] Wyn Galbraith: Like de Boom is the tree too.
[15:01] Jody Huet: thank you very much, Joe
[15:01] Joe Linden: thanks for being a good crowd for this (was prepared for more of a treatment) ;-)
[15:01] Wyn Galbraith: We're engaged in SL so only burn Lindens after the meeting.
[15:01] Mm Alder: Bye
[15:01] Wyn Galbraith: LOL
[15:01] Wyn Galbraith: Thanks Rob.
[15:01] Q Linden: zero, got a few minutes
[15:01] Wyn Galbraith: You do that so well.
[15:01] Jason Swain: yes, A very good talk, Thank you Joe
[15:01] Q Linden: ?
[15:01] Rob Linden: thanks everyone!
[15:02] Baba Yamamoto: ;0
[15:02] Joe Linden: / take care all
[15:02] Zero Linden: Uhm, sure
[15:02] Zero Linden: sure Q
[15:02] Squirrel Wood: http://www.dream.at/temp/ohours.htm