Hard Light Productions Forums
Modding, Mission Design, and Coding => FS2 Open Coding - The Source Code Project (SCP) => Topic started by: miskat on December 13, 2006, 11:41:51 am
-
I've just enabled -voicer to try to get it working (using the same built-in OfficeXP recognition software, all trained up and working alright) but for some reason, the game does not respond to commands given. I try things like "All fighters disable my target" and get no response.
Wordpad and MSWord dictation works alright, but nothing in game. Is it broken in 3.6.9rc7dot9 or did I do something wrong/miss an install step somewhere along the way?
-
*sighs* I just read that it isn't IN the 3.6.9rc builds... Darn... just got everything working all spiffy too, now I have to switch builds and find one that does it all. =(
-
Since I'm in a good mood.
http://fs2source.warpcore.org/exes/karajorma/20061213-369CVS.rar
-
This is 3.6.9 with Voice recognition? CVS?
I must say, I've always been pretty confused about what the difference was between CVS builds and official builds was. =( And honestly, I have no idea where this came from. >.< *sighs and feels ignorant* I probably missed something somewhere.
At any rate, thank's a ton, Kara.
PS. The date stamped on it... 12/13/2k6? That's today... did you just build this? O.O
PSS. Damn... voice recognition doesn't work in the build that's in your link. I get two error messages at startup "failed to create windows message" and "failed to init voice recognition" I get neither error and voice recognition works fine in the build that madprof posted toward the end of the voice recognition thread in the recent build forum.
-
I never use voice so it's possible I didn't set the options correct. I'll take a look.
-
Alrighty. I remember people mentioning the problem when the first voice recognition build was released, but I never saw a solution, it just magically went away or something. lol
I really appreciate the help. I don't use wingmen otherwise... can't think+move my hands fast enough, I guess. =( But with voice, it's easy, and really really fun! =D
-
Me appreciates the build as well. It should help me get my techroom entries right... :lol:
:nervous:
-
Nope. Looks like I built it right. I guess the bug is back.
-
Grarg. XD Alrighty, time to go diving. I WILL find out why this is happening if it's the last thing I do!
XD
Thanks for all your help, Kara.
-
Kara... I really feel bad asking for this, but could you point me to a place where I could download the copy of voicerec.cpp you used to compile your build? I would love to compare it to the copy that MadProf uploaded along with his build. Hopefully I wil lbe able to spot something. ^^;;;
Then again... that might be very presumptuous of me. BUT... I think I found the problem (or at least what could be causing it... and that MAY not be code-based at all, but the code is helping me understand it which means I might be able to fix it... anyway.)
If not, then no problem. ^.^
Thanks again!
-
You can grab the entire codebase by following the advice here (http://www.hard-light.net/forums/index.php/topic,41234.msg842224.html#msg842224).
Or you can compare the differences between two versions of the code here (http://fs2source.warpcore.org/cgi-bin/cvsweb/cvsweb.cgi/fs2_open/code/sound/voicerec.cpp)
-
Kara, you're a god.
I haven't found EXACTLY what's causing the problem yet, but I know the problem is that, for some reason, the SR won't identify Freespace2 as a target window for events.
The code looks the same between the working build and the broken build when I looked at the chunk that handles initilization of the SR engine, so it must be an anomaly on the user end of things, but what could cause the SR engine to fail like that is beyond me. The only thing I could think of is that the SR engine was being initilized before Freespace2 had wormed it's way into windows accepting it as a valid window (after all, when I get the error messages, there isn't actually a window yet.)
This is when I wish I knew more about how graphical C++ programing in Windows really worked. lol
Thanks again for all your help, Kara, it's been mightily appreciated. And now I can get the codebase and really pour over this! I might learn something! XD
Thanks again!
-
Ugh. The last time this happened was because somebody messed with the project file. I hope that didn't happen again.
Actually, I hope it did, because that would make the bug easy to find. :)
Is this in Mantis?
-
No I have not mantis'd it. I'm still not 100% sure this isn't an error on my end, Windows being picky or something. If I should Mantis it, I will, though. Just figured I'd try to really be sure first. XD
-
I get it too. Stick it in Mantis.
-
It's been Mantis'd.
-
I added my theory to the Mantis bug, so see if that helps it at all.
-
http://fs2source.warpcore.org/exes/karajorma/20061215-SpeechRecognitionBuild.rar
Try that. Should work again.
-
Your new build does indeed work, Kara. Thanks, I owe you. ...I owe all of you, actually. lol
Just seems to be missing the improvements that MadProf made (not requireing a pause, and added commands) but I'm going to see if I can't deal with those myself. XD Gotta figure out how all this works. =P
Is the phrase.xml file a madular extension of the code, so that the available commands can be changed on the fly, or is it a file that is used when compiling in order to work that phrase list into the resulting .exe?
A modular phrase list would be downright awesome, but lord knows we'd all find a way to screw that up! XD
-
Not my work. Taylors. I just happened to have a copy of MSVC open when I noticed his comment and realised it would take about a minute to check if he was right about the cause.
Mad Prof's stuff is missing because the build is a 3.6.9 branch build and his stuff is in the HEAD branch. I'll commit this to CVS after a little bit more testing and then you can simply grab one of RedMenace's CVS builds to use it.
-
Sounds great. I just followed your instructions to access the CVS repository and get the code on my own PC.
Time to have a little fun. XD
EDIT: Looking through voicer.cpp I realized how silly I was asking about on the fly editing of the available phrases. XD Might be able to on-the-fly change what phrases that the recognition engine will recognize, but you certainly couldn't tell the engine what to DO with those phrases on the fly. =( Not without a total overhaul of the engine, and Lord knows how icky that could be. XD
-
I was hoping a few other people would test it and see if it worked first.
Anyway I'll add it to CVS later today either way
-
Just found the "Does this build have speech option" in the Features/Troubleshoot tab.
This build does have speech, but when I launch a new session, the intro movie flickers over the SCP background rather than the background going away.
-
It's not so much as the Experimental->voice recognition feature that needs testing.