speech2text-development Mailing List for simon
Brought to you by:
speechtotext
You can subscribe to this list here.
| 2010 |
Jan
|
Feb
|
Mar
|
Apr
|
May
|
Jun
(1) |
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2011 |
Jan
|
Feb
|
Mar
|
Apr
(15) |
May
(26) |
Jun
(62) |
Jul
(36) |
Aug
(36) |
Sep
(16) |
Oct
(12) |
Nov
(10) |
Dec
(18) |
| 2012 |
Jan
(5) |
Feb
(3) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Peter G. <gr...@si...> - 2012-02-03 15:25:42
|
Hi! As part of the migration to the KDE infrastructure, I'd like to move the mailing list discussions to kde...@kd.... Please subscribe to the new list here: https://mail.kde.org/mailman/listinfo/kde-accessibility The list is (currently) very, very low traffic - let's change that :) Best regards, Peter |
|
From: Thomas W. <tho...@gm...> - 2012-02-03 00:34:47
|
Hi, Agreed. given the fact this list doesn't get a lot of list traffic I think just talking about Simon here would be perfectly fine. Plus I'm on a number of mailing lists anyway and would prefer to be able to keep up with projects like this all in one place rather than having to sub to a new list to keep track of development of projects like this that improve the accessibility of the KDE desktop. On 2/2/12, Jeremy Whiting <jpw...@kd...> wrote: > Peter, > > Welcome to kde a11y :) As you may or may not have noticed, > kde...@kd... doesn't get much if any traffic from users > besides once in a while a question or two. Mostly we use it for > development, so I would suggest just using this list for simon > development. I'm not sure what others think, but I think it would be > cool to follow how simon develops (/me ashamedly admits he hasn't > tried simon in a long time...) > > BR, > Jeremy > |
|
From: Jeremy W. <jpw...@kd...> - 2012-02-02 22:49:06
|
Peter, Welcome to kde a11y :) As you may or may not have noticed, kde...@kd... doesn't get much if any traffic from users besides once in a while a question or two. Mostly we use it for development, so I would suggest just using this list for simon development. I'm not sure what others think, but I think it would be cool to follow how simon develops (/me ashamedly admits he hasn't tried simon in a long time...) BR, Jeremy On Fri, Jan 27, 2012 at 2:52 AM, Peter Grasch <gr...@si...> wrote: > Hello folk of kde...@kd..., > > as some of you might know, simon is moving to the KDE infrastructure. > > On sourceforge we had two Mailing lists: One for simon development and one for > simon users. The last one never really took off (most users prefer an online > forum over a mailing list apparently) but the development list is seeing quite > steady traffic. > > As part of the move, I'd like to migrate the existing mailing lists to KDE. > Seeing as we're basically joining the KDE accessibility team, I was wondering > if it would be appropriate to use this mailing list to communicate or if we > should apply for our own list. > > Basically, I think there are three options: > 1. Use kde-accessibility for simon development discussions > 2. Create a new mailing list for simon development > 3. Create a new mailing list kde-accessibility-devel for simon, jovie, etc. > development > > Personally, I'd prefer #3 but I'd like to get some feedback of the lists > subscribers and of course KDE accessibility developers: What do you think? > > Best regards, > Peter > _______________________________________________ > kde-accessibility mailing list > kde...@kd... > https://mail.kde.org/mailman/listinfo/kde-accessibility |
|
From: Peter G. <gr...@si...> - 2012-01-27 09:55:56
|
Hello folk of kde...@kd..., as some of you might know, simon is moving to the KDE infrastructure. On sourceforge we had two Mailing lists: One for simon development and one for simon users. The last one never really took off (most users prefer an online forum over a mailing list apparently) but the development list is seeing quite steady traffic. As part of the move, I'd like to migrate the existing mailing lists to KDE. Seeing as we're basically joining the KDE accessibility team, I was wondering if it would be appropriate to use this mailing list to communicate or if we should apply for our own list. Basically, I think there are three options: 1. Use kde-accessibility for simon development discussions 2. Create a new mailing list for simon development 3. Create a new mailing list kde-accessibility-devel for simon, jovie, etc. development Personally, I'd prefer #3 but I'd like to get some feedback of the lists subscribers and of course KDE accessibility developers: What do you think? Best regards, Peter |
|
From: Peter G. <gr...@si...> - 2012-01-18 08:08:42
|
On Sunday 15 January 2012 18:29:16 Frederik Gladhorn wrote: > Congratulations! > This is great news and makes me super happy :) > > Btw, at the Ubuntu Developer Summit, I spoke to some of our KDE packagers > and there will be Simon packages for the next Ubuntu release [1]. Awesome news! Thanks! Best regards, Peter |
|
From: Frederik G. <gla...@kd...> - 2012-01-15 17:29:24
|
Congratulations! This is great news and makes me super happy :) Btw, at the Ubuntu Developer Summit, I spoke to some of our KDE packagers and there will be Simon packages for the next Ubuntu release [1]. Cheers Frederik [1] https://blueprints.launchpad.net/ubuntu/+spec/desktop-p-kubuntu-accessibility On 15. jan. 2012, at 12.25, Peter Grasch wrote: > On Saturday 14 January 2012 23:00:51 Peter Grasch wrote: >> The plan is to push the code to the KDE repo as soon as possible (I actually >> wanted to do it today but now have to wait for the q&a hooks to be disabled >> for the initial push) to start the review process. After the code has >> landed we'll gradually migrate the current development and testing stations >> from the sourceforge repo to the KDE repo. > Okay, the initial push has been completed. > > As part of the process, some repository were split' and some merged so that > the current codebase looks like this: > > simon: Application suite > https://projects.kde.org/projects/playground/accessibility/simon > Code: > git clone gi...@gi...:simon (rw) > git clone git://anongit.kde.org/simon (ro) > > simon-tools: Additional tools from the Benefit and Astromobile Project: > speechcal, panic, simonstatus, simonskype, simone and simon-touch. > Code: > git clone gi...@gi...:simon-tools (rw) > git clone git://anongit.kde.org/simon-tools (ro) > > Please setup your local repositories accordingly. > > If you do end-up committing in our old sf- or github repository, we can > extract the commit later so don't worry. If you can't do it on your own, just > drop me a line and I'll help you. > > Best regards, > Peter > > ------------------------------------------------------------------------------ > RSA(R) Conference 2012 > Mar 27 - Feb 2 > Save $400 by Jan. 27 > Register now! > http://p.sf.net/sfu/rsa-sfdev2dev2 > _______________________________________________ > Speech2text-development mailing list > Spe...@li... > https://lists.sourceforge.net/lists/listinfo/speech2text-development |
|
From: Peter G. <gr...@si...> - 2012-01-15 11:28:44
|
On Saturday 14 January 2012 23:00:51 Peter Grasch wrote: > The plan is to push the code to the KDE repo as soon as possible (I actually > wanted to do it today but now have to wait for the q&a hooks to be disabled > for the initial push) to start the review process. After the code has > landed we'll gradually migrate the current development and testing stations > from the sourceforge repo to the KDE repo. Okay, the initial push has been completed. As part of the process, some repository were split' and some merged so that the current codebase looks like this: simon: Application suite https://projects.kde.org/projects/playground/accessibility/simon Code: git clone gi...@gi...:simon (rw) git clone git://anongit.kde.org/simon (ro) simon-tools: Additional tools from the Benefit and Astromobile Project: speechcal, panic, simonstatus, simonskype, simone and simon-touch. Code: git clone gi...@gi...:simon-tools (rw) git clone git://anongit.kde.org/simon-tools (ro) Please setup your local repositories accordingly. If you do end-up committing in our old sf- or github repository, we can extract the commit later so don't worry. If you can't do it on your own, just drop me a line and I'll help you. Best regards, Peter |
|
From: Peter G. <gr...@si...> - 2012-01-14 22:06:31
|
Hello, simon developers :) IMPORTANT: Please read at least the note about the branches at the bottom. We're currently trying to migrate the simon source repository (and afterwards also the forum, bug tracking, etc.) to the KDE infrastructure. Two new repositories have already been created: simon: Containing simon, simond, ksimond, afaras, sam, ssc and sscd - the core suite. simon-tools: Containing additional programs and tools like simon-touch, speechcal, simonskype, etc. The plan is to push the code to the KDE repo as soon as possible (I actually wanted to do it today but now have to wait for the q&a hooks to be disabled for the initial push) to start the review process. After the code has landed we'll gradually migrate the current development and testing stations from the sourceforge repo to the KDE repo. In order not to interrupt the current workflow too much, we're planning to keep the sourceforge server active and supported until around the middle of February. Of course, I'll encourage everyone to use the KDE infrastructure as soon as it becomes available but because of gits decentralized nature an occasional push shouldn't be that problematic. Adam: Please apply for a KDE developer account (note your involvement in simon and put my name down as reference to speed up the process): http://community.kde.org/Sysadmin/GitKdeOrgManual#How_to_get_read- write_developer_access IMPORTANT: As preperation for the push, I cleaned up some of the old branches and finally merged windows back to sound and sound back to master. The reasoning: We only waited for the sound branch to be merged to master because it was lacking Windows support. Now that Patrick von Reth has added that in the windows branch without regressions to the Linux version it was time for an updated master. Please update your local repository and commit new changes in master or create an appropriate branch if you think it's necessary. Best regards, Peter |
|
From: Frederik G. <gla...@kd...> - 2011-12-30 10:12:10
|
Onsdag 28. desember 2011 12.54.10 skrev Peter Grasch: > Hi! > > I noticed that you are committing to the windows branch right now. On > purpose? It seemed to be the active branch. I made one compile fix and forgot to change back. > Please do keep general commits in the sound branch and reserve the windows > branch for commits needed to get the Windows version running. Will do. > And, yeah you're not being as sneaky as you think -.- Sorry. > > I'm reverting your case change. Sure. Again, sorry for the trouble. > If we change that now it'd mean we have to change it _everyhwere_: Logo, > blog, homepage, flyers, posters, reports, letterheads, polo shirts, etc. > Are you prepared to do that? I guess not, since I won't handle any of that. Cheers Frederik > Please don't try to change our whole corporate identity - at least not > without asking... > > Best regards, > Peter |
|
From: Peter G. <gr...@si...> - 2011-12-30 09:25:59
|
On Friday 30 December 2011 10:13:52 you wrote: > Onsdag 28. desember 2011 12.54.10 skrev Peter Grasch: > > I noticed that you are committing to the windows branch right now. On > > purpose? > > It seemed to be the active branch. I made one compile fix and forgot to > change back. > > > Please do keep general commits in the sound branch and reserve the > > windows branch for commits needed to get the Windows version running. > > Will do. Thanks! > > And, yeah you're not being as sneaky as you think -.- > Sorry. > > > I'm reverting your case change. > > Sure. Again, sorry for the trouble. No problem. > > If we change that now it'd mean we have to change it _everyhwere_: Logo, > > blog, homepage, flyers, posters, reports, letterheads, polo shirts, etc. > > Are you prepared to do that? > > I guess not, since I won't handle any of that. You know if I'd have to pick a name again, I'd name it "Simon" too. I understand you intention and I'm not disagreeing. It's just a huge undertaking that would require a lot of coordinated effort of a lot of different people. And the end result wouldn't help our target user group one tiny bit. So yeah, it should be done - but it's not worth the trouble right now. Best regards, Peter Ps.: Onsdag? desember? So you finally started learning Norwegian, huh? :) |
|
From: Peter G. <gr...@si...> - 2011-12-28 12:52:07
|
Hi! I noticed that you are committing to the windows branch right now. On purpose? Please do keep general commits in the sound branch and reserve the windows branch for commits needed to get the Windows version running. And, yeah you're not being as sneaky as you think -.- I'm reverting your case change. If we change that now it'd mean we have to change it _everyhwere_: Logo, blog, homepage, flyers, posters, reports, letterheads, polo shirts, etc. Are you prepared to do that? Please don't try to change our whole corporate identity - at least not without asking... Best regards, Peter |
|
From: Frederik G. <gla...@kd...> - 2011-12-26 20:42:47
|
On 23. des. 2011, at 17.26, Peter Grasch wrote: > Hi Frederik, > > On Friday 23 December 2011 13:17:11 Frederik Gladhorn wrote: >> I would really really love to see Simon hosted on the kde git servers. >> They are faster than sourceforge, work with sensible ssh keys (maybe I just >> messed up my sourceforge setup) and more people would hopefully poke at the >> code. > Yes, I agree. > > But as the move would be preceded by some time in review, I'd like to clean up > some stuff before the move (the original plan was to release 0.4 and then > switch). I am not completely sure how the review is handled at the moment. I think we gave up on the "review" part of svn since it doesn't make sense any more. If I'm not mistaken, we would invite people to review, maybe already simply using the kde git server. For now Simon could live in extragear and not be released with the main KDE release yet. Since the move is technically not much work, I think it won't interrupt development at all. It should give a big boost to translations once established. > Seeing as 0.4 would probably still take a couple of months given the rough > state of the current codebase and exams in January (I'm looking at Februrary / > March for a possible release): Does this still make sense in your opinion? Sounds like getting the windows/sound/master branch united sounds like a good plan. Once that is done, I'm for a release asap - release early, release often ;) I will ask KDE sysadmins for their opinion wrt moving simon. Greetings Frederik >> What is the status of this? Getting kde git accounts for existing >> contributors will be easy. > Yes, only a couple of people have accounts on the current SF infrastructure. > >> Happy "whatever it is you celebrate" ;) > Happy holidays! > > Best regards, > Peter > > ------------------------------------------------------------------------------ > Write once. Port to many. > Get the SDK and tools to simplify cross-platform app development. Create > new or port existing apps to sell to consumers worldwide. Explore the > Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join > http://p.sf.net/sfu/intel-appdev > _______________________________________________ > Speech2text-development mailing list > Spe...@li... > https://lists.sourceforge.net/lists/listinfo/speech2text-development |
|
From: Peter G. <gr...@si...> - 2011-12-23 16:33:57
|
On Friday 23 December 2011 14:29:46 Frederik Gladhorn wrote: > Hello all :) > looking at the simon git repo... I would propose to update the master > branch. Currently the windows branch gets most updates, I don't see a > reason why this shouldn't be the master branch. The windows branch is split off the sound branch and implements the necessary fixes and changes to get simon running on Windows again. As soon as this is done and doesn't break anything else, it'll be merged back to the sound branch. The sound branch is de-facto the master branch because I would like a master branch that works on all plattforms (didn't think it'd take _that_ long when I implemented the ALSA backend). As soon as the Windows fixes are merged into the sound branch, I'll merge sound back to master. > Another thing would like to see is adding tags for the sha1s that get put > into a release. Good idea! Best regards, Peter |
|
From: Peter G. <gr...@si...> - 2011-12-23 16:29:41
|
Hi Frederik, On Friday 23 December 2011 13:17:11 Frederik Gladhorn wrote: > I would really really love to see Simon hosted on the kde git servers. > They are faster than sourceforge, work with sensible ssh keys (maybe I just > messed up my sourceforge setup) and more people would hopefully poke at the > code. Yes, I agree. But as the move would be preceded by some time in review, I'd like to clean up some stuff before the move (the original plan was to release 0.4 and then switch). Seeing as 0.4 would probably still take a couple of months given the rough state of the current codebase and exams in January (I'm looking at Februrary / March for a possible release): Does this still make sense in your opinion? > What is the status of this? Getting kde git accounts for existing > contributors will be easy. Yes, only a couple of people have accounts on the current SF infrastructure. > Happy "whatever it is you celebrate" ;) Happy holidays! Best regards, Peter |
|
From: Frederik G. <gla...@kd...> - 2011-12-23 13:29:54
|
Hello all :) looking at the simon git repo... I would propose to update the master branch. Currently the windows branch gets most updates, I don't see a reason why this shouldn't be the master branch. Another thing would like to see is adding tags for the sha1s that get put into a release. Greetings Frederik |
|
From: Frederik G. <gla...@kd...> - 2011-12-23 12:17:22
|
Hi, I would really really love to see Simon hosted on the kde git servers. They are faster than sourceforge, work with sensible ssh keys (maybe I just messed up my sourceforge setup) and more people would hopefully poke at the code. What is the status of this? Getting kde git accounts for existing contributors will be easy. As far as I understand the technical solution is very easy (just a git push away). Cheers Frederik Happy "whatever it is you celebrate" ;) |
|
From: Peter G. <gr...@si...> - 2011-12-20 07:06:29
|
On Monday 19 December 2011 13:59:37 Adam Nash wrote: > > Does that help? > > Yeah, thanks, it was pretty easy to diagnose with that error message. I > wasn't creating the acoustic model cache skeleton (the prompts files > with specific sample groups) if there was a static model, but I was > still using those prompts files when the language model changed. I just > made it so that it always sets up the cache skeleton whenever it gets a > new prompts file. I deleted everything in the cache and started it up > with a static model and everything worked. Great. > >> Also, make sure that the Cancel button works -> don't work on the > >> real > >> configuration but a copy! > > I'm working on making the sample group context config work with the kcm > format (I didn't know anything about it before, and I just sort of > hacked a widget together and threw it into the SoundSettings thing). I > assume that it will fall into place a little cleaner when I actually > know what's going on. The .tag and non-cpp or standard .ui files throw > me off. Okay. As always, the documentation on techbase.kde.org is quite good and you'll find plenty of examples in the simon code. But I'm sure you'll figure it out.. > As far as the singleton stuff goes, I agree that one singleton that > manages the rest would make sense. I noticed that the simonview > destructor (which deletes simoncontrol) never gets called, and there is > something in main.cpp about it being deleted automatically. I deleted > it after the program exits just to see what would happen and it crashes, > apparently because of something in scenariomanager that depended on > simonview or part of it. Should I be doing anything about this > singleton problem? It seems like a big project (although, a quick fix > seems like it would be as simple as finding dependencies between > singletons and deleting them in the right order). Yes, but you can leave that to me. I'm hoping I can take a look at that (among a couple of other long standing issues) over the holidays. Best regards, Peter |
|
From: Adam N. <ada...@gm...> - 2011-12-19 18:59:47
|
> There shouldn't be, I guess. > > I just dug up the concrete error message: > Could not adapt prompts. Does the file > "c:/users/<foo>/appdata/roaming/.kde/share/apps/simond/models/default/cached//acoustic > models/default/prompts" exist? > > It looks like simond tries to access a cached prompts file that doesn't exist. > That happened when using a static model (no training data). > > Does that help? Yeah, thanks, it was pretty easy to diagnose with that error message. I wasn't creating the acoustic model cache skeleton (the prompts files with specific sample groups) if there was a static model, but I was still using those prompts files when the language model changed. I just made it so that it always sets up the cache skeleton whenever it gets a new prompts file. I deleted everything in the cache and started it up with a static model and everything worked. >> Also, make sure that the Cancel button works -> don't work on the real >> configuration but a copy! I'm working on making the sample group context config work with the kcm format (I didn't know anything about it before, and I just sort of hacked a widget together and threw it into the SoundSettings thing). I assume that it will fall into place a little cleaner when I actually know what's going on. The .tag and non-cpp or standard .ui files throw me off. As far as the singleton stuff goes, I agree that one singleton that manages the rest would make sense. I noticed that the simonview destructor (which deletes simoncontrol) never gets called, and there is something in main.cpp about it being deleted automatically. I deleted it after the program exits just to see what would happen and it crashes, apparently because of something in scenariomanager that depended on simonview or part of it. Should I be doing anything about this singleton problem? It seems like a big project (although, a quick fix seems like it would be as simple as finding dependencies between singletons and deleting them in the right order). -Adam |
|
From: Peter G. <gr...@si...> - 2011-12-17 13:48:39
|
On Friday 16 December 2011 11:54:41 Frederik Gladhorn wrote: > > Frederik suggested to look into K_GLOBAL_STATIC and it seems to provide > > a > > means to fix that problem: > > http://api.kde.org/4.x-api/kdelibs- > > apidocs/kdecore/html/group__KDEMacros.html#ga75ca0c60b03dc5e4f9427263bf4 > > 043c7 > > > > I haven't yet looked into it properly, though. > > I don't think that is a good solution. It's something I tried for a bit > until I realized that there > are so many different singletons and such. > > Having two (and more) singletons depend on each other makes it super > hard to untangle the mess. > For a single singleton I would suggest K/Q_GLOBAL_STATIC. But that is > more to make it thread safe, since the simple static variables are not > and you could end up with two instances (one that never gets used). I > don't think this actually happens in simon. I didn't read the code > really, so this is what I remember from a month ago or so. > > Without thinking about it too much, I would try to reduce the singletons > to one. That one would create all other objects that are singleton-like > at the moment. Its sole purpose would be management of the other classes. +1 Best regards, Peter |
|
From: Frederik G. <gla...@kd...> - 2011-12-16 10:54:55
|
On 12/14/2011 06:40 PM, Peter Grasch wrote: > Hi Adam, > > On Tuesday 13 December 2011 19:02:01 Adam Nash wrote: <snip> >> Speaking of ContextManager - I noticed that its instance never gets >> deleted and its destructor is never called because the destructor in >> SimonControl is never called. This also means that none of the >> conditions are deleted. This is probably bad I think? :P >> I am not on good terms with the singleton gods, so I will be happy to >> take advice on how to deal with this. Well, if the singletons need to be around until the app is closed it is kind of fine. I still don't like it much since its quite confusing right now. The problem with not deleting things is usually leaks so that the application consumes more and more memory while it is running. That should not be the problem with the singletons. > If you find out, tell me about it :) > > No, seriously that's something that's already causing problems in other parts > of simon. > > Frederik suggested to look into K_GLOBAL_STATIC and it seems to provide a > means to fix that problem: > http://api.kde.org/4.x-api/kdelibs- > apidocs/kdecore/html/group__KDEMacros.html#ga75ca0c60b03dc5e4f9427263bf4043c7 > > I haven't yet looked into it properly, though. I don't think that is a good solution. It's something I tried for a bit until I realized that there are so many different singletons and such. Having two (and more) singletons depend on each other makes it super hard to untangle the mess. For a single singleton I would suggest K/Q_GLOBAL_STATIC. But that is more to make it thread safe, since the simple static variables are not and you could end up with two instances (one that never gets used). I don't think this actually happens in simon. I didn't read the code really, so this is what I remember from a month ago or so. Without thinking about it too much, I would try to reduce the singletons to one. That one would create all other objects that are singleton-like at the moment. Its sole purpose would be management of the other classes. Cheers Frederik > >>> Btw.: There seems to be a very annoying bug somewhere in the caching >>> system. If you start with a fresh installation, simon will complain >>> about not finding something in the cache. Could you please have a look? >> I will look at this. I thought that if there was no cache it just sets >> a new one up (I was pretty routinely deleting my entire cache when >> working with the context adapter). Also, I don't think I made anything >> that pops up error messages, just kDebug output. Will there be a >> difference between a new installation and an old one with just the >> context cache deleted? > There shouldn't be, I guess. > > I just dug up the concrete error message: > Could not adapt prompts. Does the file > "c:/users/<foo>/appdata/roaming/.kde/share/apps/simond/models/default/cached//acoustic > models/default/prompts" exist? > > It looks like simond tries to access a cached prompts file that doesn't exist. > That happened when using a static model (no training data). > > Does that help? > > Best regards, > Peter > > ------------------------------------------------------------------------------ > Cloud Computing - Latest Buzzword or a Glimpse of the Future? > This paper surveys cloud computing today: What are the benefits? > Why are businesses embracing it? What are its payoffs and pitfalls? > http://www.accelacomm.com/jaw/sdnl/114/51425149/ > _______________________________________________ > Speech2text-development mailing list > Spe...@li... > https://lists.sourceforge.net/lists/listinfo/speech2text-development |
|
From: Peter G. <gr...@si...> - 2011-12-14 17:43:14
|
Hi Adam, On Tuesday 13 December 2011 19:02:01 Adam Nash wrote: > Ok, I'm done with jury duty now. I found the state of new york guilty > of wasting my time :D :) > I made the existing sample groups available in a combobox in the > samplegroupcontext table view. Ok, cool. > > Also, make sure that the Cancel button works -> don't work on the real > > configuration but a copy! > > I will get to this next. Maybe I can make it make direct changes (I > can't copy ContextManager without angering the gods of the singleton > pattern beyond reason), but when ok or apply is clicked it saves them? > And if cancel is clicked it will just load its saved configuration. > That sounds easy to me, but is it too easy to be a good idea? Hm. That sounds like a bit of a hack to me. I can think of two - imho cleaner - solutions for this: * Make the actual configuration a separate, copyable object that is held and managed by the ContextManager singleton. That way you can copy it and apply / reset becomes straight forward. * Or: If copying the configuration would be expensive (which it likely is, considering that there are possibly a lot of plugins involved), you can just keep track of a list of "scheduled changes" and apply / reset that list on save() / cancel(). (hint: Command pattern) > Speaking of ContextManager - I noticed that its instance never gets > deleted and its destructor is never called because the destructor in > SimonControl is never called. This also means that none of the > conditions are deleted. This is probably bad I think? :P > I am not on good terms with the singleton gods, so I will be happy to > take advice on how to deal with this. If you find out, tell me about it :) No, seriously that's something that's already causing problems in other parts of simon. Frederik suggested to look into K_GLOBAL_STATIC and it seems to provide a means to fix that problem: http://api.kde.org/4.x-api/kdelibs- apidocs/kdecore/html/group__KDEMacros.html#ga75ca0c60b03dc5e4f9427263bf4043c7 I haven't yet looked into it properly, though. > > Btw.: There seems to be a very annoying bug somewhere in the caching > > system. If you start with a fresh installation, simon will complain > > about not finding something in the cache. Could you please have a look? > > I will look at this. I thought that if there was no cache it just sets > a new one up (I was pretty routinely deleting my entire cache when > working with the context adapter). Also, I don't think I made anything > that pops up error messages, just kDebug output. Will there be a > difference between a new installation and an old one with just the > context cache deleted? There shouldn't be, I guess. I just dug up the concrete error message: Could not adapt prompts. Does the file "c:/users/<foo>/appdata/roaming/.kde/share/apps/simond/models/default/cached//acoustic models/default/prompts" exist? It looks like simond tries to access a cached prompts file that doesn't exist. That happened when using a static model (no training data). Does that help? Best regards, Peter |
|
From: Adam N. <ada...@gm...> - 2011-12-14 00:02:13
|
Ok, I'm done with jury duty now. I found the state of new york guilty of wasting my time :D I made the existing sample groups available in a combobox in the samplegroupcontext table view. > As this now in the configuration screen of simon, you get the save() and load() > calls from KCModule. > I'd load the config once on program start (and just retrieve the already loaded > config in load()) and save (and apply) it whenever save() is called. I guess I forgot to read the second half of your last email, I did something that was probably dumber - I will try to switch it to this way after I get to the following stuff. > Also, make sure that the Cancel button works -> don't work on the real > configuration but a copy! I will get to this next. Maybe I can make it make direct changes (I can't copy ContextManager without angering the gods of the singleton pattern beyond reason), but when ok or apply is clicked it saves them? And if cancel is clicked it will just load its saved configuration. That sounds easy to me, but is it too easy to be a good idea? Speaking of ContextManager - I noticed that its instance never gets deleted and its destructor is never called because the destructor in SimonControl is never called. This also means that none of the conditions are deleted. This is probably bad I think? :P I am not on good terms with the singleton gods, so I will be happy to take advice on how to deal with this. > > Btw.: There seems to be a very annoying bug somewhere in the caching system. > If you start with a fresh installation, simon will complain about not finding > something in the cache. Could you please have a look? > > I will look at this. I thought that if there was no cache it just sets a new one up (I was pretty routinely deleting my entire cache when working with the context adapter). Also, I don't think I made anything that pops up error messages, just kDebug output. Will there be a difference between a new installation and an old one with just the context cache deleted? -Adam |
|
From: Peter G. <gr...@si...> - 2011-12-07 07:59:57
|
Hi Adam, On Tuesday 06 December 2011 18:36:11 Adam Nash wrote: > Sorry for the delay, I have been having jury duty. :D > When I just compiled the acoustic model with all of the scenarios, but > did not change the language model to the "all scenarios are active" > default I got this error: > > " > The word "START" is used in your training-samples but is not contained > in your wordlist. > > What do you want to do? > " > > I assume that it won't let you make that generic acoustic model if the > generic language model is not also present? Anyways, I just made it so > that every time the acoustic model is recompiled, the generic language > model is temporarily set as the language model. Yes, that's what I'd have recommended. But I'd say there was (and maybe is) a bug in there somewhere because the model compilation adapter should normally strip away training data for untrained words before starting the compilation to avoid this issue. But compiling the model with the full lexicon is of course the recommended way. Just make sure it still works if you add a word and don't train it (independent of the scenario- / acoustic model switching). > Also, I seem to have the context-driven sample group changes working, Great! Can't wait to try this out :D > [...] but this error keeps popping up, even though everything seems to be > acting as expected: > > "The model synchronization reported the following error: > Could not complete Synchronization > " This is weird but it's not a common problem. Could it be that restoring a model from the cache doesn't lock the temporary path or something like that? If the synchronization does not complete it normally means that SynchronisationManager::commit() fails... > One more thing - I asked about how to store the > samplegroup conditions when the program is closed- how would you > recommend I do that? I asume that it will be loaded up when > contextmanager starts and then saved when forced to save and/or when > contextmanager is destroyed. As this now in the configuration screen of simon, you get the save() and load() calls from KCModule. I'd load the config once on program start (and just retrieve the already loaded config in load()) and save (and apply) it whenever save() is called. Also, make sure that the Cancel button works -> don't work on the real configuration but a copy! Btw.: There seems to be a very annoying bug somewhere in the caching system. If you start with a fresh installation, simon will complain about not finding something in the cache. Could you please have a look? Best regards, Peter |
|
From: Adam N. <ada...@gm...> - 2011-12-06 23:36:23
|
Hey Peter, Sorry for the delay, I have been having jury duty. Here are some things I ran into: When I just compiled the acoustic model with all of the scenarios, but did not change the language model to the "all scenarios are active" default I got this error: " The word "START" is used in your training-samples but is not contained in your wordlist. What do you want to do? " I assume that it won't let you make that generic acoustic model if the generic language model is not also present? Anyways, I just made it so that every time the acoustic model is recompiled, the generic language model is temporarily set as the language model. Also, I seem to have the context-driven sample group changes working, but this error keeps popping up, even though everything seems to be acting as expected: " The model synchronization reported the following error: Could not complete Synchronization " I will be trying to fix the error box thing as well as improve the usability in the mean time, but any input you have on the error box would be appreciated because I don't want to just hack around it because everything seems to be working - I assume something must be wrong and I just can't tell what. One more thing - I asked about how to store the samplegroup conditions when the program is closed- how would you recommend I do that? I asume that it will be loaded up when contextmanager starts and then saved when forced to save and/or when contextmanager is destroyed. -Adam |
|
From: Peter G. <gr...@si...> - 2011-12-02 08:40:28
|
Hi Adam, On Tuesday 29 November 2011 20:50:26 Adam Nash wrote: > I finished the stuff that lets the user specify the conditions that will > trigger certain sample groups/acoustic models. I put it in the > recording sub-section of the simon config menu - let me know if you want > it elsewhere. Also, let me know of any usability tweaks that you think > should be done. The recording settings are getting a bit crowded but I can't think of any place more suitable either. Usability: IMHO it would be nice if one could enter the sample group in the add-condition dialog. Also - and I think this is really important - the sample group selection should be a drop-down filled with: a.) The sample groups of the current sound configuration b.) The sample groups of the prompts c.) The previously configured sample group But the context thing is clearly only going to be set up by power users anyway... > I got and error message about the monophones being unavailable when the > language model context was changing. According to this debug output I > guess that the file was there, it was just being used elsewhere. I > don't think that this happened before: I can't really test it now as I don't have a valid model here but it was a trivial bug -> Likely fixed. > Also I'm getting this error even though I have training samples for all > of my vocabulary: > > "Could not start recognition because the system reports that the > recognition is not ready. > Please check if you have defined a vocabulary, an appropriate grammar > and recorded a few trainings samples. > The system will then, upon synchronization, generate the model which > will be used for the recognition." Well this is to be expected if your model generation fails. Basically this is the error for "no active model yet". > I have it so that the sample groups will be sent to the server now when > they change, but I have not been able to test it because of these > errors. They may be due to something that got mucked on my system, so I > will try to do some troubleshooting. Nope, sorry those were upstream fixes that I've probably introduced when I fixed a couple of other issues with models that only contain single words for our upcoming Italian speech model. Best regards, Peter |