20101218
Meeting started by UndiFineD at 23:02:18 UTC. The full logs are available at http://mootbot.libertus.co.uk/speechcontrol/2010/speechcontrol.2010-12-17-23.02.log.html .
Meeting summary
LINK: http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition (UndiFineD, 23:19:13) LINK: getting Simon on board would be a great help, as to whether it is possible, that is different matter. I do, however think it is worth a try. (UndiFineD, 23:29:32) LINK: https://blueprints.launchpad.net/ubuntu/+spec/ubuntu-mobile-voice-user-interface (UndiFineD, 23:29:47) ACTION: RainCT patch julius with bedahr (UndiFineD, 23:56:30) ACTION: waywardgeek looks at speeds speech with openMary (UndiFineD, 00:28:48) LINK: http://www.cyber-byte.at/wiki/index.php/Licensing (UndiFineD, 00:37:05) LINK: http://www.cyber-byte.at/wiki/index.php/Licensing (UndiFineD, 00:37:10)
Meeting ended at 01:05:20 UTC.
Votes
JackyAlcine for Development Coordinator
- For: 6 Against: 0 Abstained: 0
Action items
- RainCT patch julius with bedahr
- waywardgeek looks at speeds speech with openMary
Action items, by person
- bedahr
- * RainCT patch julius with bedahr
- RainCT
- * RainCT patch julius with bedahr
- waywardgeek
- * waywardgeek looks at speeds speech with openMary
People present (lines said)
- UndiFineD (135)
- bedahr (97)
JackyAlcine (88)
- waywardgeek (65)
- phillw (47)
- hajour (38)
AlanBell (23)
ChrisDruif (17)
- meetingology (16)
DragonEyes (9)
- RainCT (7)
- pedro3005 (4)
- ibuclaw (2)
- Cheri703 (1)
Full Log
23:02:18 <UndiFineD> #startmeeting
23:02:18 <meetingology> Meeting started Fri Dec 17 23:02:18 2010 UTC. The chair is UndiFineD. Information about MeetBot at http://wiki.ubuntu.com/AlanBell.
23:02:18 <meetingology> Useful Commands: #topic #action #link #idea #voters #vote #chair #action #agreed #help #info #endmeeting.
23:02:33 <waywardgeek> bedahr: Hi, Peter. Good to see you.
23:02:47 <UndiFineD> I would like to welcome you all, it is an honour
23:03:05 <UndiFineD> hajour is our leader
23:03:17 <UndiFineD> but she is not a very fast typer
23:04:11 <UndiFineD> We have quite some challenges ahead
23:04:30 <UndiFineD> and we still have to decide on a full mission statement
23:04:48 <UndiFineD> I like what you wrote JackyAlcine
23:05:07 <JackyAlcine>
Thank you.
23:05:46 <UndiFineD> so maybe it is best to first introduce ourselves
23:06:09 <phillw> ;phillw
23:06:11 <DragonEyes> Hi, (https://wiki.ubuntu.com/phillw) is an admin person, wiki stuff, tidying up stuff creating new entries. Passionate about Lubuntu (https://wiki.ubuntu.com/Lubuntu), Accessibility (https://wiki.ubuntu.com/Accessibility/Team) and UBT (https://wiki.ubuntu.com/BeginnersTeam) Also familiar with LAMP and web stuff.
23:06:22 <JackyAlcine> ;JackyAlcine
23:06:27 <DragonEyes> the up-coming developer from New York that plans to bring about a new wave of computer science and information processing as we know yet. (https://wiki.ubuntu.com/jackyalcine http://jdevelopthis.blogspot.com)
23:06:38 <UndiFineD> I am UndiFineD (Keimpe de Jong) and I am supporting hajour in this project mostly
23:06:45 <UndiFineD> ;UndiFineD
23:06:46 <DragonEyes> (https://wiki.ubuntu.com/UndiFineD) now belongs to pedro (https://wiki.ubuntu.com/pedro3005), please offer him every assistance. (His name is Keimpe)
23:07:13 <bedahr> I don't think this will work for me so here I go: I am bedahr (Peter Grasch) and mainly work on the open source speech recognition solution simon
23:07:49 <UndiFineD> ;hajour
23:07:54 <DragonEyes> Manuèla (https://wiki.ubuntu.com/hajour), who is bringing in a whole load of accessibility knowledge to the (L)ubuntu project. Take the time to read her poems. phillw (https://wiki.ubuntu.com/phillw) is proud to be her Mentor as a padawan for UBT (see ;padawan or https://wiki.ubuntu.com/BeginnersTeam).
23:08:05 <AlanBell> I am AlanBell, an Ubuntu member and contributor to the Accessibility team
23:08:15 <ChrisDruif> I am ChrisDruif. I'm mostly here out of general interest. Hope to be off help, but I'm not sure I can
23:09:18 * RainCT is Siegfried Gevatter, Ubuntu/Debian/Zeitgeist Developer (maintaining the Julius package, for instance) and writing stupid stuff like espeak-gui :p
23:10:00 <waywardgeek> waywardgeek: aka Bill Cox. Big geek for waaay too long, losing central vision and motivated to help build tools I need to remain a Linux geek indefinately.
23:10:00 <hajour> hai i am hajour leader of the team.my was told it was hard to build a program like this also impossible.i thought it was possible.s i started this program 11dec 2010 and started to look for all the people wo ever have worked on the programs and i stil do to join. now we here to share the knowledge together on this first meeting
23:10:19 <phillw> UndiFineD: I think that we can safely say, we have people in several teams, and I'd also like put on record our thanks for you showing interest is this team.
23:10:28 <UndiFineD> Yorvyk ?
23:10:35 * JackyAlcine claps in recognition.
23:10:53 <UndiFineD> if only that would work :P
23:11:41 <UndiFineD> We had some initial idea's like this:
23:11:41 <UndiFineD> Keep effort to a minimum, re-use existing technology
23:11:41 <UndiFineD> Solve potential issues
23:11:41 <UndiFineD> Build an Accessible application to preform tasks by speech (or normal input)
23:11:41 <UndiFineD> Have the results returned to the user (with information overload protection)
23:11:42 <UndiFineD> Long term: startrek like communications, dictate a document
23:12:10 <UndiFineD> but that is not a real mission ..
23:12:42 <waywardgeek> UndifineD: I call it a "coversational interface". Don't know if that's the right term, but I want to talk and listen to my computer.
23:12:50 <UndiFineD> as JackyAlcine stated: Sorting out the programmers and logicians behind SpeechControl; flushing out the needed goals to accomplish, figuring out what has been done already and what to do from here.
23:13:22 <UndiFineD> waywardgeek, that is the intention, but there more than one disabilities
23:13:42 <UndiFineD> or handicaps
23:13:55 <waywardgeek> UndiFinD: I think you'll find some people are very familiar with the current state. Luke, for example. I'm not too bad myself, but behind the guys who do this full time.
23:14:23 <bedahr> is Luke here?
23:14:47 <AlanBell> no
23:14:54 <AlanBell> his nick is themuso
23:14:59 <hajour> i dont think so bedahr
23:15:13 <UndiFineD> so, we have a couple of works (programs) and I would like to see our efforts to be synchronized, and potential problems solved
23:15:13 <hajour> i have invite him
23:15:18 <bedahr> ok... I thought it might have been Yorvyk...
23:15:40 <ibuclaw> Just thought I might chime in: http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition
23:16:04 <ibuclaw> I recall reading it back in May, it's likely still relevant today. =)
23:16:35 <UndiFineD> initially, with the least amount of effort gain as much as possible
23:16:48 <UndiFineD> RainCT, mentioned a deadline
23:17:21 <UndiFineD> "The date to remember for Natty is Feb 24, that's feature freeze, so any new application should be in the archive by then, I'd suggest that he join the Debian accessibility team and try to push the package through, if there's no progress by mid january, feel free to ping RainCT and we'll take another look at itThe date to remember for Natty is Feb 24, that's feature freeze, so any new application should be in the archive by
23:17:21 <UndiFineD> then, I'd suggest that he join the Debian accessibility team and try to push the package through, if there's no progress by mid january, feel free to ping RainCT and we'll take another look at it"
23:17:34 <UndiFineD> oops double paste
23:17:35 <phillw> bedahr: Yorvyk is steve. He's a Lubuteer, but also does get involved on other channesl. / teams.
23:18:23 <hajour> we can look at the link from ibuclaw later i think.
23:18:33 <UndiFineD> next to this we have another 'dead' project on ubuntu: Voice Driven UI
23:18:36 <bedahr> phillw: thanks
23:18:51 <JackyAlcine> I'm examining it in a side window now, hajour; I'll PM (if needed) you some information on it as we go.
23:18:54 <waywardgeek> ibuclaw: There are a ton of stories to tell about why accessibility progress has stalled in critical areas. It's been something of a blood bath
23:18:54 <UndiFineD> which has a blueprint which we might be able to take up upon
23:19:13 <UndiFineD> #link http://robertfortner.posterous.com/the-unrecognized-death-of-speech-recognition
23:19:20 <JackyAlcine> True, waywardgeek (+1 for wishful thinking)
23:20:51 <JackyAlcine> * glances about.
23:20:55 <UndiFineD> so with a little time pressure (2 months) I call for actionable items
23:20:58 <ChrisDruif> Shouldn't meetingology confirm that a link has been added?
23:21:10 <waywardgeek> So, here's a couple points. I believe there are enough VI programmers out there to make incredible progress, but we don't have them organized. The full time positions are too few to complete the job.
23:21:12 <hajour> yes
23:21:46 <AlanBell> ChrisDruif: it is quite a quiet bot
23:21:46 <waywardgeek> Also, the organizations that currently lead accessibility are all handicapped in some ways.
23:21:58 <phillw> As we are firstly seeking voice control of the Apps, gettig full dictation is the Holy Grail.
23:22:19 <phillw> ChrisDruif: it has been added
23:22:24 <JackyAlcine> phillw, that's a large-term goal; keep in mind.
23:22:29 <ChrisDruif> Thanks phillw
23:22:34 <waywardgeek> For example, FSF/GNU has an excellent effort, but they aren't able to endorse emacspeak, the most productive programming environment for the blind.
23:23:13 <waywardgeek> We've also got the Gnome team, but they have little influence over the GTK+ team and aren't able to help KDE, for example.
23:23:20 <phillw> Is why we must focus on what is currently possible.
23:23:27 <JackyAlcine> Don't mind me asking, waywardgeek, is it a licensing disambiguation with emacspeak?
23:23:29 <waywardgeek> I believe it is all posible.
23:24:17 <ChrisDruif> But possibly not at this moment waywardgeek
23:24:21 <waywardgeek> T.V. Raman uses commercial TTS, and for good reason: it makes him more productive. Since emacspeak encourages users to use commercial TTS, it can't be supported by FSF.
23:25:23 <waywardgeek> ChrisDruif: Not only is it all possible, though not instantly, I'm willing to tackle what I feel is the hardest code. That's why I'm diving into SR and TTS.
23:25:35 <UndiFineD> ok, what i read here is all good, things to think about, but I want action items for the next to months, what do you think is considered possible ?
23:26:12 <AlanBell> is there an agenda for this meeting?
23:26:19 <UndiFineD> bedahr, is it possible to port back the patches from your julius back into it, for example, so dabien likes you better
23:26:22 <waywardgeek> Improving the install process for Ubuntu 11.04, making it more accessible, would be a fine goal.
23:26:29 <UndiFineD> AlanBell, topic
23:26:31 <JackyAlcine> >_<, that's what I forgot, AlanBell. Not really.
23:26:34 <bedahr> UndiFineD: yes, probably
23:26:49 <AlanBell> waywardgeek: that is out of the scope of this project
23:26:57 <bedahr> I don't think this is the only problem tough concerning simon and licenses
23:26:59 <AlanBell> waywardgeek: but very much in scope of the Accessibility team
23:27:34 <bedahr> even tough it is not needed for usual use cases simon can use the HTK - which isn't free software
23:27:37 <waywardgeek> AlanBell: Sorry for being a noob to the team... what is the scope of this project?
23:27:39 <AlanBell> waywardgeek: however the focus for this cycle is getting unity accessibility fixed, there might be a little work on the installer if there is time
23:27:50 <bedahr> so I don't expect the die hard debians to ever be really happy with the package to be honest
23:28:02 <UndiFineD> the scope / mission has not been fully determined, which we should
23:28:07 <AlanBell> waywardgeek: speech recognition and text to speech and integrating them together and with the desktop
23:28:15 <phillw> I think getting Simon on board would be a great help, as to whether it is possible, that is different matter. I do, however think it is worth a try.
23:28:28 <JackyAlcine> Then waywardgeek is on the right track, AlanBell.
23:28:58 <bedahr> I mean I am of course heavily biased but I know that you could get pretty far quite fast by using simon
23:28:58 <JackyAlcine> I think that Simon has to have a bit of more friendly API
23:29:02 <waywardgeek> Getting Simon included would be outstanding, even with initial limitations.
23:29:06 <UndiFineD> second to this, I would like us to take on the blueprint: https://blueprints.launchpad.net/ubuntu/+spec/ubuntu-mobile-voice-user-interface
23:29:08 <JackyAlcine> so then it'd be easier to port from desktop environments.
23:29:20 <bedahr> JackyAlcine: what exactly are you looking for?
23:29:32 <UndiFineD> #link getting Simon on board would be a great help, as to whether it is possible, that is different matter. I do, however think it is worth a try.
23:29:32 <UndiFineD> <JackyAlcine> Then waywardgeek is on the right track, AlanBell.
23:29:32 <UndiFineD> <bedahr> I mean I am of course heavily
23:29:47 <UndiFineD> #link https://blueprints.launchpad.net/ubuntu/+spec/ubuntu-mobile-voice-user-interface
23:30:16 <phillw> AlanBell: sorry, the rough agenda is at http://openetherpad.org/tGnASiT5le
23:31:05 <bedahr> maybe we could do a rough breakdown and compare the status quo with the inital prototype targeted for Ubuntu 11.04?
23:32:27 <JackyAlcine> bedahr, if you have more of a native, dare I say, set up for Simon, like simon, and then gsimon, and ksimon
23:32:35 <phillw> bedahr: +1, I really do think that 11.1o is a realistic goal, there is so much going on and with an 8 week deadline, I really cannot see it happening in time.
23:32:59 <waywardgeek> bedahr: How is the Debian packaging coming? Is there a version I can put on Launchpad.net as a source package?
23:33:19 <hajour> AlanBell, we have notify you on the etherpad earlier
23:33:29 <bedahr> ok then lets focus on 11.10 from the start; Given we have unlimited resources, what would be the goal then?
23:33:40 <bedahr> waywardgeek: There is a complete debian package of simon 0.3.0 on my launchpad account
23:34:21 <waywardgeek> bedahr: sweet. I'll contact you by e-mail for getting it to some users for testing.
23:34:30 <JackyAlcine> I feel that since it's orientated for KDE; GTK+/GNOME developers are going to be reluctant to interact with it.
23:34:37 <bedahr> JackyAlcine: We do provide a server / client infrastructure that does the "real work". The server links to KDE for the moment but just because I didn't have time to strip it everywhere; Porting it to Qt alone should be trivial
23:35:08 <phillw> bedahr: something that companies with millions of dollars cannot acheve, we have to stay within the realms of what is possible.
23:35:13 <AlanBell> I don't think the KDE dependency is a deal breaker
23:35:19 <JackyAlcine> In that case, I guess so then bedahr.
23:35:23 <bedahr> so you could write gsimon which is just another simon client; but I wouldn't really focus on it at the beginning to be honest...
23:35:45 <waywardgeek> bedahr: I agree
23:36:00 <AlanBell> current version works, just isn't in the repos
23:36:12 <UndiFineD> and that is an issue
23:36:13 <bedahr> phillw: true, but I'd suggest we go a bit crazy to get an overview of the "vision". Then we can start to be realistic and get a clear cut mission of what is actually possible. But thats just a suggestion, obviously
23:36:14 <AlanBell> we can use it to develop scenarios
23:36:40 <phillw> as I and a couple are from Lubuntu, lxde would be nice
23:37:04 <AlanBell> the scenarios could then run on the current simon, a minimal gtk simon front end to simond or a full gnome simon client
23:37:21 <AlanBell> but we don't need to wait before developing scenarios
23:37:29 <phillw> I can go and quietly nag, for our devs to look further into it.
23:37:54 <UndiFineD> bedahr, yes that is what we want, and if we stick to the deadline and our initial goal is to obtain the blueprint, that would help us a lot in the future, as we would be able to have an UDS session about it
23:38:21 <AlanBell> UndiFineD: what do you mean by "obtain the blueprint"?
23:38:33 <UndiFineD> make use of it
23:38:46 <phillw> AlanBell: how long to the UDS- in budhappest?
23:39:04 <AlanBell> UndiFineD: we can just start implementing it
23:39:21 <AlanBell> phillw: next UDS is May 8th I think
23:39:44 <JackyAlcine> I've began making my own scenario for Simon, it's supposed to interact with Pidgin via D-Bus.
23:39:48 <bedahr> hm one problem: Scenarios involve the simon client; The actual plugins will run on the client (to launch applications, etc.)
23:40:07 <phillw> That's a more realistic time scale than 14th Feb.
23:40:12 <bedahr> so the gsimon client will differ from the ksimon client scenario wise
23:40:12 <UndiFineD> for those who haven't, please read that blueprint, as it could help us in some sponsoring
23:40:16 <bedahr> this shouldn't happen
23:40:42 <bedahr> we should really make the plugin interface client agnostic; But this would involve restructuring simon itself a bit
23:41:13 <JackyAlcine> How much work/elbow grease would that take, bedahr ?
23:41:14 <bedahr> JackyAlcine: If you use the git version there is a dbus plugin for that
23:41:27 <phillw> it's why I suggested using python and either c / C++.
23:41:45 <bedahr> Hard to say really as I fully rely on the KDE plugin infrastructure; There are also plugins who interact with other parts of simon which are ofcourse KDE-styled as well
23:41:45 <UndiFineD> phillw, the feb 24th deadline is to getting new packages in ubuntu
23:42:04 <UndiFineD> like simon
23:42:05 <phillw> piy, as I'm not a programmer these days, I can only throw ideas in to discuss.
23:42:08 <bedahr> phillw: that wouldn't really change anything; With python I'd have used pykde4
23:42:58 <JackyAlcine> bedahr, yes, but it's faulty.
23:43:00 <AlanBell> personally I am not bothered about desktop interfaces and toolkits and the religious wars about which to use
23:43:06 <RainCT> bedahr: Can you give a quick summary of what features Simon (backend, not gui) provides?
23:43:19 <bedahr> ok getting back to the mission, I'm still not quite certain of what should be achieved in time for 10.10; Voice controlling of applications is definitely a priority I guess. What about speech output?
23:43:37 <phillw> bedahr: I'd have to check with the lubuntu devs, but as far as I know, they have not had problems with kde stuff.
23:43:56 <bedahr> AlanBell: Agreed; Qt / KDE4 was just the pragmatically best choice for us when we started (past experience, etc.)
23:44:03 <UndiFineD> bedahr, that would be speech dispatcher
23:44:10 <bedahr> RainCT: sure;
23:44:36 <AlanBell> it would be nice to get some additional speech engines nicely integrated with speech dispatcher
23:44:41 <AlanBell> such as openMary
23:44:47 <bedahr> RainCT: the backend is a TCP server with simple (sqlite based) user management; Multiple users can connect to one simond client; The server manages their speech model and the actual recognition
23:44:54 <JackyAlcine> AlanBell, I've been looking at that.
23:44:58 <UndiFineD> yes AlanBell
23:44:59 <bedahr> AlanBell: libsimontts provides that :P
23:45:48 <AlanBell> bedahr: does that help other apps that use speech dispatcher like orca?
23:45:53 <bedahr> RainCT: the client just opens a TCP Stream to the server and comunicates with a custom protocol; The client is responsible for capturing audio from the input device and segmenting it (detecting silences); He then sends the raw sound data to the server which returns a recognition result
23:45:56 <bedahr> (basically)
23:46:34 <bedahr> AlanBell: no, it's just our internal abstraction layer for our tts output stuff; But you theoretically use it from orca as well, of course
23:46:56 <RainCT> OK, I see
23:47:00 <bedahr> at the moment it wraps speech-dispatcher, OpenMARY and plain wave files
23:47:33 <AlanBell> bedahr: yeah, I read about that, cool stuff!
23:47:41 <bedahr> the tts stuff is mainly for our dialog system (http://www.youtube.com/watch?v=3_RTJsq9m4I)
23:47:51 <RainCT> bedahr: By the way, you said it should to put your julius patches into the Debian package, right? Will see if I can take a look at it during holidays
23:48:31 <bedahr> yes I said that this would be a possibility...
23:48:49 <bedahr> you'd probably have to clean them up a bit tough I guess
23:49:04 <bedahr> we removed some stuff we don't need, etc.
23:49:42 <bedahr> Also we use cmake as build system now to keep in line with the rest of simon; You'd have to adapt Julius build system to build dynamic libraries as well (atm it only builds static ones)
23:50:25 <RainCT> Yeah, that's something I want to do anyway before I get it into Debian.
23:50:38 <bedahr> great 23:51:28 * RainCT is off to bed - cya
23:51:34 <UndiFineD> ok so that is something you could action ?
23:51:38 <bedahr> RainCT: bye
23:51:49 <hajour> sleep wel RainCT
23:51:52 <phillw> thanks RainCT
23:52:05 * JackyAlcine salutes RainCT
23:52:06 <ChrisDruif> Thanks and good night RainCT
23:52:15 <UndiFineD> thank you RainCT
23:52:21 <hajour> and thanks RainCT
23:53:00 <RainCT> thanks
23:53:13 <phillw> ;bye | RainCT
23:53:15 <DragonEyes> RainCT: take care and ensure the penguin does not catch you
23:55:01 <hajour> bedahr, so the endconlusion what will be done .can you tell that
23:55:09 <UndiFineD> ok so we have some ideas
23:55:34 <UndiFineD> I hope RainCT and bedahr can work julius stuff out
23:55:39 <bedahr> hajour: To be honest I'm still trying to get an overview of the project
23:55:48 <ChrisDruif> So RainCT is going to patch julius? That would help simon get into the repositories...
23:56:05 <bedahr> yes that would probably be step one
23:56:30 <bedahr> longer term I think we should think about finally porting simon to sphinx;
23:56:30 <UndiFineD> #action RainCT patch julius with bedahr 23:56:30 * meetingology RainCT patch julius with bedahr
23:56:58 <hajour> maybe to take a 10 minutes to pause
23:57:29 <JackyAlcine> yes, bedahr, but like CMU Sphinx needs voices
23:57:39 <bedahr> Julius does too
23:57:47 <bedahr> and there is a model converter HTK -> SPHINX
23:58:05 <JackyAlcine> No, its standard voices are horrible.
23:58:19 <JackyAlcine> but openMary's German Spike sounds pretty good.
23:58:41 <waywardgeek> Can someone tell me where to find install instructions for openMary?
23:58:53 <bedahr> waywardgeek: just download and run the .jar file
23:59:05 <bedahr> the java installer is very easy and even downloads voices on the fly
23:59:16 <hajour> attention :maybe it's good to have a short pause of 10 minutes
23:59:28 <JackyAlcine> lol, bedahr took me 20 mins to figure that out.
23:59:31 <UndiFineD> I think people would like to have both possibilities
23:59:32 <JackyAlcine> and okay, hajour.
23:59:40 <waywardgeek> bedahr: Just tried that... "java openmary-standalone-install-4.1.1.jar crashes"
00:00:03 <bedahr> waywardgeek: hm... worked for me
00:00:20 <waywardgeek> Getting further now... chmod +x on the jar file seems to help
00:00:25 <bedahr>
00:00:34 <bedahr> ok I think we have a break now?
00:00:50 <UndiFineD> and also speech is not the only thing we are thinking of...
00:01:17 <bedahr> break over?
00:01:24 <bedahr> UndiFineD: What do you mean?
00:01:49 <JackyAlcine> recognition is what you're addressing UndiFineD .
00:01:49 <UndiFineD> making use of speechdispatcher braille is also a wish
00:02:03 <bedahr> oh ok
00:02:26 <hajour> ok now break to let it in al the information
00:02:36 <waywardgeek> Is no one working on openMary support in SD?
00:03:11 <hajour> thhen all can think it over
00:04:03 <hajour> 10 minutes
00:04:23 * AlanBell goes to bed, night all o/
00:04:30 <hajour> and maybe get a drink ore something
00:04:32 <UndiFineD> good night AlanBell
00:04:39 <hajour> good night AlanBell
00:04:44 <bedahr> AlanBell: good night
00:04:49 <waywardgeek> lage
00:04:52 <waywardgeek> later
00:08:30 <phillw> UndiFineD: Luke had it running in 9.10, http://www.mail-archive.com/ubuntu-accessibility@lists.ubuntu.com/msg03740.html He'd be the one to ask.
00:08:45 <phillw> thanks for your time AlanBell
00:08:55 <UndiFineD> Ok so now we an idea what is going on, and have wishes
00:08:55 <UndiFineD> what should fit our mission ?
00:08:55 <UndiFineD> can we get programs like simon and sonic into debian / ubuntu by the 24th ?
00:08:55 <UndiFineD> are there any remaining licencing issues?
00:08:55 <UndiFineD> next meeting:
00:09:11 <ChrisDruif> So....to get this right: not only do we want to make speech/voice control available in linux.....but we want to also convert text to braille?
00:09:21 <JackyAlcine> I'm looking into openMary, waywardgeek.
00:09:44 <JackyAlcine> And I sent an e-mail asking if I could re implement the code in C++.
00:09:50 <hajour> ChrisDruif, its a wish.that means if it is possible
00:10:01 <JackyAlcine> And possibly ChrisDruif, it's one of those +5 things.
00:10:09 <UndiFineD> ChrisDruif, see the pre-meeting notes on what do we want to accomplish?
00:10:28 <phillw> ChrisDruif: there has already been been work done on that, it is a target. But we need to priotise.
00:10:30 <ChrisDruif> Well...it should be a small step from tts to ttb...
00:10:51 <bedahr> Are we building a single application or are we trying to bring voice features to the default desktop?
00:11:08 <UndiFineD> ChrisDruif, the speech dispatcher already offers it
00:11:32 <UndiFineD> there are multiple disabilities and many combinations
00:11:38 <JackyAlcine> bedahr, to be honest; I'm not sure anymore.
00:11:49 <UndiFineD> What do we want to accomplish?
00:11:49 <UndiFineD> Command
00:11:49 <UndiFineD> Execute
00:11:49 <UndiFineD> Return results
00:11:49 <UndiFineD> Optional: Manage reptitive tasks (e-mail, scheduling, etc.) for enduser.
00:12:10 <bedahr> well what exactly does "Return results" mean?
00:12:14 <bedahr> return from where?
00:12:29 <UndiFineD> bedahr, yes a single application for this all would be best, as it solves all issues people have
00:12:45 <UndiFineD> except for note taking
00:13:29 <UndiFineD> bedahr, startrek like: computer home many people are in this room ? there are 22 people in this room ?
00:13:51 <phillw> ChrisDruif: it is all about scavenging about live, and resurrecting / re-invigourating teams that have given up. As ibuclaw said earlier, the road is littered with failed projects. We have to keep realistic short term goals, whilst aiming towards what is currently not fully possible.
00:14:22 <ChrisDruif> bedahr: it means that StarTrek communication; when you say for instance "Open email"...the computer responds "Opening email" or "You have 192 unread emails"
00:14:44 <UndiFineD> 192 ... that must be me
00:14:59 <bedahr> ChrisDruif: That I know; I was asking about the "Return results" point in the list; Or is this the star trek thing?
00:15:12 <UndiFineD> therefore we came to artificial intelligence
00:15:24 <UndiFineD> as that gives interaction
00:15:34 <ChrisDruif> UndiFineD: same minds come together
00:15:50 <UndiFineD> yup, we made sure of that
00:16:25 <JackyAlcine> bedahr, "Return results" in the sense that desktop events and manipulations can be done with the desktop.
00:16:34 <bedahr> ah ok
00:16:51 <bedahr> so we will probably want to integrate with at-spi and the like
00:16:55 <JackyAlcine> That room counting feature can be done via some OpenGL..
00:17:01 <JackyAlcine> Yup.
00:17:06 <UndiFineD>
00:18:13 <UndiFineD> waywardgeek, also made something, without even listening, really cool, speed adjustment for the output voices, thruh sonic
00:18:24 <bedahr> ok so summing up we'd have an application listening to voice input that is capable of launching executables (configurable), manages active windows / applications (at-spi) and provides simple dialogs (basic "ai")
00:18:52 <bedahr> right?
00:18:59 <UndiFineD> exactly
00:19:26 <bedahr> oh and there will be a tts component as well - probably as separate entity (?) as we will probably just launch / depend on orca (?)
00:19:33 <bedahr> (or Jovie on KDE obviously)
00:20:12 <ChrisDruif> Might be a good idea...
00:20:14 <UndiFineD> bedahr, I would depend on speechdispatcher as orca is too
00:20:44 <UndiFineD> and that gives direct control over outputs
00:21:22 <UndiFineD> waywardgeek, would you like to say something about sonic ?
00:21:22 <bedahr> yes but do we need direct control?
00:21:24 <JackyAlcine> bedahr, yes.
00:21:44 <JackyAlcine> And if possible, it can narrate virtually any and every application on the desktop.
00:21:50 <ChrisDruif> I know simon isn't appropriate for dictation....maybe a separate application is better at it and we should just execute it with simon
00:21:52 <bedahr> what do we really need from the tts system that e.g. orca couldn't provide?
00:22:24 <waywardgeek> Hi, guys... I'm back. Just had to feed the kids 00:22:27 * hajour gives the voice on waywardgeek fot to tell about his program
00:22:44 <hajour> wb waywardgeek
00:22:50 <phillw> bedahr: there are people who hate orca voices with a passion
00:22:50 <bedahr> ChrisDruif: It that separate dictation could do dictation, so could probably simon... The problem lies with free speech models not with the software (mostly)
00:22:59 <JackyAlcine> waywardgeek, you made sonic?
00:23:29 <waywardgeek> Yep. It's a simple app to speed up speech.
00:23:45 <bedahr> phillw: but orca just uses speech-dispatcher which supports a wide variety of backends, right? Even if we implemented our own tts system atop of speech-dispatcher we would share the same voices...
00:23:50 <waywardgeek> It works pretty well when integrated into the TTS engine. I'm just looking at Mary now.
00:24:23 <JackyAlcine> yeah bedahr
00:24:48 <JackyAlcine> that's why we could aim for attaching ourselves to SpeechDispatcher.
00:24:49 <waywardgeek> While speech-dispatcher does, IMO, leave a lot to be desired in terms of ease of developing TTS interfaces, it works.
00:25:08 <phillw> bedahr: RainCT: did some good work with espeak... http://forum.phillw.net/viewtopic.php?f=14&t=33 I do think it is worth having a look at. 00:25:18 * hajour people can we let talk waywardgeek
00:25:35 <waywardgeek> * waywardgeek may not have much to say
00:25:46 <JackyAlcine> I'm curious, how does sonic work?
00:25:55 * JackyAlcine looks for source code and binaries to download.
00:25:58 <UndiFineD> well 6x speed up aint nothing ...
00:26:06 <hajour> maybe you can answer questions
00:26:18 <UndiFineD> does it slow down too waywardgeek ?
00:26:22 <waywardgeek> There's no magic. All that's different is I had speed listeners involved, so the algorithms work well at far beyond 2X speed up.
00:26:30 <waywardgeek> It slows down and also changes pitch.
00:27:16 <waywardgeek> It's only purpose is to enable commercial and FOSS TTS engines, as well as audio book apps, support speeds required by blind speed listeners.
00:27:33 <JackyAlcine> hmmm.
00:27:34 <JackyAlcine> Okay.
00:27:39 <waywardgeek> So, I'll check out openMarry. If it sucks at high speed like most of the others, I can fix that
00:28:07 <UndiFineD> wonderful
00:28:34 <ChrisDruif> Alright everyone...I'm off....tired as hell and tomorrow I'm working on my own project again together with my buddy
00:28:36 <bedahr> Well the problem with openMary imho is - while the quality (at least for German voices) is excellent - it is really, really slow and being Java has some hefty dependencies as well
00:28:48 <UndiFineD> #action waywardgeek looks at speeds speech with openMary 00:28:48 * meetingology waywardgeek looks at speeds speech with openMary
00:28:52 <phillw> by the way UndiFineD have you endedmeeting yet? :P
00:28:55 <bedahr> ChrisDruif: ok good night
00:29:03 <UndiFineD> nope
00:29:11 <ChrisDruif> Thanks bedahr
00:29:13 <JackyAlcine> night, ChrisDruif ; keep me posted!
00:29:13 <JackyAlcine>
00:29:18 <JackyAlcine> and use the webkit!
00:29:22 <hajour> goodnight ChrisDruif
00:29:23 <JackyAlcine> ill explain later.
00:29:24 <UndiFineD>
00:29:34 <ChrisDruif> Alright, will do
00:29:35 <UndiFineD> good night ChrisDruif
00:29:39 <phillw> ;bye | ChrisDruif
00:29:40 <ChrisDruif> Webkit for what?
00:29:40 <DragonEyes> ChrisDruif: take care and ensure the penguin does not catch you
00:29:52 <JackyAlcine> webkit for HTML rendering.
00:29:59 <JackyAlcine> instead of the generic isht Pidgin uses.
00:30:02 <JackyAlcine> more cleaner.
00:30:04 <JackyAlcine> but yeah.
00:30:15 <JackyAlcine> I'll explain tomorroww
00:30:19 <UndiFineD> ok next: what should fit our mission ?
00:30:21 <ChrisDruif> Aha....I'll talk about it with my buddy
00:30:32 <JackyAlcine> Probably next Tuesday.
00:30:38 <JackyAlcine> I was thinking Friday.
00:30:45 <JackyAlcine> but that seems busy..
00:30:50 <JackyAlcine> isn't it Christmas then?
00:30:50 <hajour> hajour
00:31:18 <waywardgeek> Ha! I thought those HMM voices sounded MBROLA-ish.
00:31:37 * JackyAlcine is confused by waywardgeek's discovery.
00:31:49 <hajour> note UndiFineD> ok next: what should fit our mission ?
00:31:52 <bedahr> waywardgeek: what? really? I don't know about the english ones but the German ones are awesome
00:32:14 <waywardgeek> Some MBROLA voices are quite good
00:32:34 <waywardgeek> But you can hear the distortion from the MBE/LPC encoding
00:32:48 <bedahr> I tried those too (again, German) and they don't hold a candle to bits3-hmm
00:33:05 <bedahr> actually, come to think of it I do know the English hmm voices and think they are quite good as well
00:33:12 <UndiFineD> can our mission be to extend simon to become a full accessibilty application ?
00:33:21 <bedahr> hey I'd like that
00:33:33 <UndiFineD> it seems logical
00:34:00 <bedahr> yeah it certainly provides a lot of what you are looking for
00:34:15 <bedahr> also, it should be quite easily changable
00:34:19 <UndiFineD> but is that compatible with lubuntu phillw ?
00:34:28 <phillw> UndiFineD: they would (simon) have to clarify the licence, and then it would be up to a MOTU to propose it.
00:34:46 <JackyAlcine> who had a problem installing openMary?
00:34:56 <JackyAlcine> Remember to chmod 755 it!
00:34:58 <bedahr> phillw: what problems do you see with the license?
00:35:13 <bedahr> or in other words: What should we do to "clarify" it?
00:35:27 <phillw> UndiFineD: as long as is not gnome, it is not impossible.
00:36:14 <UndiFineD> ok phillw
00:37:05 <UndiFineD> http://www.cyber-byte.at/wiki/index.php/Licensing
00:37:10 <UndiFineD> #link http://www.cyber-byte.at/wiki/index.php/Licensing
00:37:38 <phillw> bedahr: http://www.voxforge.org/home/about/comments/simon-and-htk-licensing
00:38:37 <bedahr> phillw: yes, if you read the full article at lwn.net (http://lwn.net/Articles/348267/) you can find quite a good discussion about simon and surrounding licenses
00:38:39 <phillw> I recall it bringing been brought to my attention before, I believe Debian blocked it's adoption because of the ambiguities.
00:38:51 <bedahr> this lead to the wiki article linked above
00:38:52 <UndiFineD> that would mean each user, that needs HTK, would have to read the htk license agree to it and downlad it separate from within the application
00:39:14 <JackyAlcine> Why is that a problem?
00:39:16 <bedahr> yes using the HTK on the clients is definitely not an option
00:39:34 <UndiFineD> unless done aas I just described bedahr
00:39:52 <JackyAlcine> Wait, download ?? separately from the application?
00:39:58 <phillw> I'm not even sure if they would agree to it in resrticted extras.
00:40:01 <JackyAlcine> I want to understand this aspect..
00:40:05 <UndiFineD> yup, that is in the htk license
00:40:23 <bedahr> guys, don't forget that we really shouldn't need the HTK for what we are planning here
00:40:33 <UndiFineD> yes, that would be best
00:41:00 <waywardgeek> I think bedahr has already well covered the licence issues. I was possibly the first person to get freaked out by it. The answer to me is that short term you only ship HHM models, and long term we build a more complete solution.
00:41:02 <phillw> if simon does not need htk, it is possible to get it through in time.
00:41:43 <UndiFineD> phillw, why is that ?
00:41:57 <bedahr> before we try to force people to install the HTK I say we port the simon training to SPHINX. I haven't had the time to start with it yet but if this project takes of this should certainly be doable...
00:42:24 <waywardgeek> right
00:42:38 <UndiFineD> sphinx is nice too, seen the vedics application screencasts ?
00:42:41 <JackyAlcine> **HMM models.
00:43:01 <phillw> UndiFineD: GPL Vs Restrictive licences with non GPL licences, FOS does like them.
00:43:03 <UndiFineD> but sphinx 4 also has license issues i think ...
00:43:19 <phillw> UndiFineD: are they all GPL?
00:43:21 <waywardgeek> Here's my take on Simon. It's the only speech control application in Linux that is even close to usable. We need something usable ASAP.
00:43:39 <waywardgeek> We can refine over time, but we need to have something out there.
00:44:01 <phillw> waywardgeek: then we need to get hold of Luke and see if he can provide a fully GPL version.
00:44:19 <bedahr> phillw: SPHINX is BSD as far as I know
00:44:34 <JackyAlcine> it is =/
00:44:38 <waywardgeek> Actually, I think bedahr is the expert there...
00:44:46 <bedahr> huh?
00:45:13 <UndiFineD> phillw, sphinx makes use of the bsd license
00:45:19 <waywardgeek> bedahr, have you had time to factor out the HTK dependencies, so we can have a Debian/Ubuntu friendly package we can ship
00:45:36 <waywardgeek> With the obvious limitations that implies.
00:45:48 <bedahr> waywardgeek: they were never factored "in". You simply select a static base model during initial setup and off you go
00:46:00 <bedahr> the HTK was never a build time dependency
00:46:00 <waywardgeek> right, then.
00:47:04 <waywardgeek> I've been off doing other stuff (like sonic), but I'll download Simon from launchpad tomorrow.
00:47:23 <UndiFineD>
00:47:36 <JackyAlcine> bedahr, do you have a CVS checkout or Bazaar for Simon?
00:47:42 <JackyAlcine> I want to look at it a bit more.
00:47:55 <bedahr> Git, yes
00:48:00 <bedahr> sf.net/projects/speech2text
00:48:00 <UndiFineD> so is sphinx a problem ?
00:48:18 <bedahr> simon doesn't use sphinx atm
00:48:36 <JackyAlcine> Sphinx's recognition is pretty bad.
00:48:46 <JackyAlcine> AlanBell and I had fun testing it out.
00:48:59 <UndiFineD> ok ...
00:48:59 <bedahr> JackyAlcine: I don't know...
00:49:10 <UndiFineD> screencasts were impresive though
00:49:22 <JackyAlcine> did you hear the voices, though?
00:49:23 <bedahr> SPHINX should be about on par with HTK/Julius
00:49:37 <JackyAlcine> I had a few friends (American and Carribean accents) try.
00:49:45 <JackyAlcine> and my mom's rough french/creole accent.
00:49:53 <UndiFineD> ok
00:49:58 <JackyAlcine> and only my friend from holland had some good results.
00:49:58 <hajour> hehe
00:49:59 <UndiFineD> how did that go
00:49:59 <waywardgeek> I tried it, too. I didn't have much luck with SPHINX.
00:50:19 <JackyAlcine> it picks up Queensland English better.
00:50:32 <JackyAlcine> Unless we build models; I'm not expecting much.
00:50:44 <UndiFineD> ok, so sphinx is not of your liking ..
00:50:54 <waywardgeek> I like th SPHINX license.
00:51:02 <bedahr> waywardgeek: me too
00:51:03 <phillw> Realistically, we have about 4 weeks to have a GPL version confirmed, else it will not be available for 11.04. As I think 11.10 is a more realistic target, we do not need to break our necks trying to rush something in.
00:51:24 <UndiFineD> true phillw
00:51:27 <JackyAlcine> True phillw
00:51:35 <bedahr> phillw: Agreed; I say we try to fix the Julius fork and see what they say
00:51:45 <JackyAlcine> But it wouldn't hurt to have an alpha.
00:51:49 <JackyAlcine> Julius-NG?
00:51:53 <waywardgeek> However, we can likely get it into Vinux 3.2, based on Ubuntu 11.04
00:52:05 <waywardgeek> And get users to start playing with it
00:52:20 <waywardgeek> That should help get it ready for Ubuntu 11.10
00:52:30 <bedahr> waywardgeek: sounds good
00:53:05 <JackyAlcine> Vinux?
00:53:07 <waywardgeek> That's the primary goal of Vinux... you wont have to jump through hoops. Guys there are anxious for anything you can give them, and they love testing.
00:53:10 <phillw> you devs should have access to their ppa's - then you can sit down and discuss which is nearest. I think Simon is at present, but it depends on how much work the others will need.
00:53:10 <JackyAlcine> ;gdefine Vinux
00:53:12 <DragonEyes> Error. The syntax for that command is: gdefine <phrase>
00:53:23 <waywardgeek> Vinux = Linux for the visually impaird, build on Ubuntu.
00:53:33 <JackyAlcine> Oh. okay.
00:53:37 <waywardgeek> Luke works pretty closely with us on various bits of code.
00:53:59 <phillw> ;add Vinux is Linux for the visually impaird, build on Ubuntu.
00:54:00 <DragonEyes> Factoid Vinux was successfully added
00:54:07 <waywardgeek> With Luke's prodding, we've been trying to expand to covering other disabilities.
00:54:26 <UndiFineD> ok next:
00:54:27 <phillw> ;add vinux is Linux for the visually impaird, build on Ubuntu.
00:54:29 <DragonEyes> Factoid vinux was successfully added
00:55:16 <waywardgeek> The main thing with Vinux is we can use it as a testing ground for unstable accessibility code.
00:55:35 <JackyAlcine> Is there a place for me to download such a system?
00:55:45 <waywardgeek> vinux.org.uk
00:55:48 <JackyAlcine> I really love virtualization, fyi.
00:55:52 <waywardgeek>
00:56:07 <hajour> i nominate JackyAlcine for Development Coordinator
00:56:22 <waywardgeek> :-D
00:56:25 <JackyAlcine> we started voting for positions already?
00:56:44 <hajour> <UndiFineD> ok next:
00:56:51 <hajour> yes
00:57:17 <hajour> phillw,
00:57:32 <hajour> voting
00:58:12 <UndiFineD> #vote JackyAlcine for Development Coordinator
00:58:12 <meetingology> Please vote on: JackyAlcine for Development Coordinator
00:58:12 <meetingology> Public votes can be registered by saying +1, +0 or -1 in channel, (private votes don't work yet, but when they do it will be by messaging the channel followed by +1/-1/+0 to me)
00:58:12 <phillw> waywardgeek: that would be excellent, I'd prefer everyone to be under the #ubuntu-accessibility banner, as I'm hopeful to get it set up as a Focus Group, which would this sort of project on a more formal footing and allow the informal team interactions officially recognised.
00:58:18 <phillw> +1
00:58:18 <meetingology> +1 received from phillw
00:58:19 <UndiFineD> +1
00:58:19 <meetingology> +1 received from UndiFineD
00:58:22 <hajour> +1
00:58:22 <meetingology> +1 received from hajour
00:58:38 <bedahr> +1
00:58:38 <meetingology> +1 received from bedahr
00:58:42 <hajour> vote s plz
00:58:43 <waywardgeek> _1
00:58:46 <waywardgeek> +1
00:58:46 <meetingology> +1 received from waywardgeek
00:58:52 <waywardgeek> correct
00:59:00 <UndiFineD> anyone else ?
00:59:01 <JackyAlcine> If only I could vote for myself.
00:59:04 <JackyAlcine> lol.
00:59:21 <JackyAlcine> there's Pendulum , jdeslaur , ibuclaw , Cheri703 and ...
00:59:26 <JackyAlcine> me.
00:59:39 <Cheri703> +1
00:59:39 <meetingology> +1 received from Cheri703
00:59:42 <UndiFineD> #voters
00:59:42 <meetingology> Current voters:
00:59:57 <UndiFineD> huh ?
01:00:02 <UndiFineD> #agrees
01:00:03 <UndiFineD> #agreed
01:00:14 <phillw> try #endvote
01:00:16 <UndiFineD> #agreed JackyAlcine for Development Coordinator
01:00:21 <UndiFineD> #endvote
01:00:21 <meetingology> Voting ended on: JackyAlcine for Development Coordinator
01:00:21 <meetingology> Votes for:6 Votes against:0 Abstentions:0
01:00:21 <meetingology> Motion carried
01:00:40 <JackyAlcine> Thank you, guys.
01:00:57 <waywardgeek> So, Ubuntu-accessibility is a good banner. I think better than GNU/FSF or Gnome.
01:00:58 <hajour> ok thats set ackyAlcine is Development Coordinator
01:01:16 <phillw> next up.... poor harrassed soul for documentation.... which would be me.
01:01:31 <UndiFineD> ok what would be a good date / time for next meeting ?
01:01:31 <waywardgeek> Congrats, JackyAlcine!
01:01:41 <hajour> lol phillw
01:01:48 <UndiFineD> phillw, that is not a question
01:01:51 <hajour> congrats JackyAlcine
01:01:59 <bedahr> JackyAlcine: congratulations!
01:02:12 <JackyAlcine> I'm for the 28th
01:02:22 <UndiFineD> congrats JackyAlcine
01:02:49 <phillw> 28th is okay with me.
01:02:54 <UndiFineD> ok we will set up a doodle up and around the 28th
01:02:55 <waywardgeek> Works here
01:03:04 <bedahr> will probably work for me too
01:03:09 <UndiFineD> # nextmeeting ?
01:03:24 <phillw> UndiFineD: all one word
01:03:40 <UndiFineD> #nextmeeting +- 28th of december
01:03:52 <UndiFineD> hey pedro3005
01:03:57 <UndiFineD> you missed it all
01:04:04 <pedro3005> hey
01:04:10 <pedro3005> yeah, but
01:04:11 <hajour> hee pedro3005
01:04:12 <pedro3005> I ate pizza
01:04:57 <UndiFineD> thank you all for coming .. we will have to leave it here, enjoy christmas and play time
01:04:58 <phillw> pedro3005: your punishment is reading two hours of logs... it was a long meeting
01:05:03 <hajour> happy Christmas
01:05:12 <hajour> lol
01:05:12 <pedro3005> hey hajour
01:05:20 <waywardgeek> merry xmas/holiday
01:05:20 <UndiFineD> #endmeeting
Generated by MeetBot 0.1.4 (http://wiki.ubuntu.com/AlanBell)
SpeechControl/Meetings/20101218 (last edited 2010-12-18 01:11:10 by D9784976)