Hard Light Productions Forums
Modding, Mission Design, and Coding => FS2 Open Coding - The Source Code Project (SCP) => Test Builds => Topic started by: RandomTiger on January 04, 2006, 01:34:03 pm
-
Hi there, this is still a bit of a work in progress but I thought you guys might want to have a look.
BEFORE RUNNING THIS YOU MUST TRAIN THE VOICE SYSTEM.
This feature is for the PC only and is activated with -voicer in the command line.
cvs isnt up to date because Im having problems connecting to it.
In XP (and possibly 2000) in control panel there should be a Speech->Speech Recognition option. Run through 'Configure mic' and 'Train Profile'. Non XP users might have to download the SAPI SDK.
The system has limitations, it only understands set commands (most in the menu) and most greek alphabeta wing names.
So no custom ship names are going to be understood.
Also running this may slow down the game as it as to spend CPU time processing your speach.
Sometimes the menu pops up with stuff it shouldnt but I've found the speech recognition to be fairly good and Im not even using a proper mic, just headphones plugged into the mic socket.
Let us know how you get on.
http://www.randomtiger.pwp.blueyonder.co.uk/freespace/fs2_open_r.zip
To be clear, this is what it should understand:
</L>
<O>wing</O>
<P VAL="0">Alpha</P>
<P VAL="1">Beta</P>
<P VAL="2">Gamma</P>
<P VAL="3">Delta</P>
<P VAL="4">Epsilon</P>
<P VAL="5">Zeta</P>
<P VAL="6">Eta</P>
<P VAL="7">Theta</P>
<P VAL="8">Iota</P>
<P VAL="9">Kappa</P>
<P VAL="10">Lambda</P>
<P VAL="11">Mu</P>
<P VAL="12">Nu</P>
<P VAL="14">Omicron</P>
<P VAL="16">Rho</P>
<P VAL="17">Sigma</P>
<P VAL="18">Tau</P>
<P VAL="19">Upsilon</P>
<P VAL="23">Omega</P>
</L>
<L PROPID="VID_Number">
<P VAL = "1">one</P>
<P VAL = "2">two</P>
<P VAL = "3">three</P>
<P VAL = "4">four</P>
<P VAL = "5">five</P>
<P VAL = "6">six</P>
<P VAL = "7">seven</P>
<P VAL = "8">eight</P>
<P VAL = "9">nine</P>
<P VAL="VID_Ships" >Ships</P>
<P VAL="VID_Wings" >Wings</P>
<P VAL="VID_AllFighters">All Fighters</P>
<P VAL="VID_AllWings" >All Wings</P>
<P VAL="VID_Reinforcements">Reinforcements</P>
<P VAL="VID_SupportShip">Support ship</P>
<P VAL="VID_SupportShip">Rearm</P>
<P VAL="VID_SupportShip">Repair</P>
<P VAL="VID_AbortSupport">Abort</P>
<P VAL="VID_Cancel">Cancel</P>
<P VAL="VID_Command">Command</P>
<P VAL="VID_DestoryTarget">Destroy my target</P>
<P VAL="VID_DisableTarget">Disable my target</P>
<P VAL="VID_DisarmTarget" >Disarm my target</P>
<P VAL="VID_DestroySubsys">Destroy subsystem</P>
<P VAL="VID_ProtectTarget">Protect my target</P>
<P VAL="VID_IgnoreTarget" >Ignore my target</P>
<P VAL="VID_FormWing" >Form on my wing</P>
<P VAL="VID_CoverMe" >Cover me</P>
<P VAL="VID_EngageEnemy" >Engage Enemy</P>
<P VAL="VID_EngageEnemy" >Attack</P>
<P VAL="VID_Depart" >Depart</P>
-
Wow, that is an awesome feature! :yes:
I'll have to search for my mic. It's a quite decent headset. :)
-
How about those of us with Dual Core Systems. Can we make the other processor handle the recognition?
-
Edit: misunderstood the question.
No idea, havent seen anything on that.
Probably not without a new SDK.
Let get it working for everyone first eh? :)
-
lol, well.
You might want to get your avatar back as well
-
Umm... my P3 OEM FS1 had the ability to do voic recognition command... :wtf:
-
Wouaw, Is it working ? We only just need to improve a little the AI behavior and we'll got a really bad ass Fs 2 :D
-
Works for me, not perfect but it works.
Now I need you guys to have a go and give me some feedback :)
-
I didn't quite understood what I should do...
The build is working fine alright, but I can't get Voice Rec. to work.
Any ideas ?
With the -Voicer enabled, I get the following error
"Failed to create the command recognition content"
And the next error message is:
"Failed to init voice rec"
-
Hi there, thanks for your feedback.
What OS are you using?
Is your mic plugged in and turned on?
If XP make sure you have trained the voice recognition system in control panel.
Otherwise you might have to install the API: http://download.microsoft.com/download/speechSDK/SDK/5.1/WXP/EN-US/speechsdk51MSM.exe
-
I'm using Windows XP
Mic is working and turned on...
Culprit could be the Voice Rec. System..
I will also install API, to see if it would be any help...
Oh, and I'm using Logitech headset with attached Mic.
-
Umm, where do you install that .exe-file ?
-
If you have XP you really shouldnt need to install anything.
Doesnt matter where you install it.
I've nevered coded anything like this before so Im not exactly going to be able to give you guru type advise on any of this.
Have you trained the voice system in control panel?
-
RandomTiger, you might just have started something tham might be impossible to stop. It may take a year or Sundays or more but I cant wait for the time where you can bark into your mic "All fighters take out those turrets, all bombers form on me and commence your bombing runs against those forward beam cannons and for God's sake get out of the way when command tells you!"
-
RandomTiger, you might just have started something tham might be impossible to stop. It may take a year or Sundays or more but I cant wait for the time where you can bark into your mic "All fighters take out those turrets, all bombers form on me and commence your bombing runs against those forward beam cannons and for God's sake get out of the way when command tells you!"
Muhahahaha :lol: But it seems it is possible to do that, UT2007 propose this kind of features.
-
Voice control is older than you think. Actually, you can already voice command your squadmates with Game Commander (http://www.gamecommander.com/)(commercial, custom engine) or Shoot (http://clans.gameclubcentral.com/shoot/)(free, MS engine).
However, having it in FSO is good because you can now link anything to a voice command with a little coding, instead of just keypresses.
-
Thats interesting knn, not having looked at this area before I didnt know about these solutions existing already. Still, I didnt spend much time on it. If you have used them yourself you could give my implementation a try and see how it measures up.
If this kind of thing becomes standard a new era of gaming will start, people like MetalDestroyer all over the world screaming orders at their computers.
-
Thats interesting knn, not having looked at this area before I didnt know about these solutions existing already. Still, I didnt spend much time on it. If you have used them yourself you could give my implementation a try and see how it measures up.
:sigh: My mic is crap, and I don't have a headset :(. Most of the time the game would be responding to the output from the speakers, probably slowing down the computer a lot.
Besides, I haven't used Shoot before, only Game Commander with ST Bridge Commander - That was cool (I had a headset then)
'Helm - Engage!' :D
What would've been even more cooler: e.g. the ability to say 'Set course for <systemname>' instead of 'Set course one' (first system in the drop down list).
If this kind of thing becomes standard a new era of gaming will start, people like MetalDestroyer all over the world screaming orders at their computers.
Snipes: 'DIVE DIVE DIVE! Hit your burners pilot'
Player: 'Down! Down! DOWN! Afterburner! I said AFTERBURNER!!'
WHAM!
;)
-
Oh, one point you dont have to bring the menu up to use these commands.
You can just say:
All fighters..... attack my target.
-
Muhahahaha But it seems it is possible to do that, UT2007 propose this kind of features.
Do UT2007 try to give a good impression with that ? (they'll need much more than that). I believe that was old history when the N64 game "Hey, you Pikachu!" was released in 2000, you can't play that game without a Microphone, it's a minimun requirement, otherwise you can't order your Pokemon to attack, defend, evade, and do other multiple moves. An intent to make the Game, as playable as the Series is, through Voice orders.
This feature is, and would be excellent for FS2/FSSCP (....if I only would have a Microphone....)
-
Plug your walkman/mp3 player headphones into the mic socket. :)
-
id test it if i had a mic.
-
This is utterly awesome.
But you know what would be even cooler? If I had a mike and headset.
:(
-
I'm sure it rocks, but it seems to be utterly impossible to use in any other Windows XP than one with English language settings. I'm just getting a message "Failed to create the command recognition context. -> Failed to init voce rec" ...
So, it doesn't work on my Finnish Windows even though it is XP (with SP2). It seems that I'd need a ******* American English version of Microsoft Office to make speech recognition work on this computer... :mad: :hopping:
...What makes me even more bitter, my Opera actually DOES recognize spoken commands quite well. But this FS2 feature seems to be dependant of Windows' own(ed) feature that is nor present in all language settings. Bad. Thing. Is there any possibility to make this feature use some other method to recognize speech? Because I don't know how the hell I'm supposed to make this thing work. All the help files considering speech recognition just keep telling me to click nonexistent buttons or entries or choose a feature that is seemingly not available. Arrrrr....
I bet this is the problem with Starfox also, being Finnish like me. Oh boy, does this suck... The API (http://download.microsoft.com/download/speechSDK/SDK/5.1/WXP/EN-US/speechsdk51MSM.exe) mentioned earlier doesn't really help at all either, it's just a bunch of files that can't be used. Unless someone knows how they can be used, they're not any use as they are, I would neet something to actually USE them...
I really hope that you people can rig this system to use some other kind of system to recognize voice commands, I really have no intention to buy an American English (or simplified/traditional chinese or japanese) version of Microsoft Office to get this working. Otherwise it would be a very nice feature, if it worked for all.
-
From microsoft.com:
Operating System Support and Requirements
Supported Operating Systems
* Microsoft Windows XP Professional or Home Edition; all language versions.
I don't know what Opera uses, it installs voice libraries when you activate voice recognition, so it's probably not M$.
You could try installing the voice SDK itself (http://download.microsoft.com/download/speechSDK/SDK/5.1/WXP/EN-US/speechsdk51.exe) instead of the redist (redistributable, the file you installed, with msm in the name), though that doesn't make much sense, but, with M$, who knows. Shoot also requires that you download the SDK, not the redist, so maybe they know something. If it still doesn't work, try Shoot (http://clans.gameclubcentral.com/shoot/) (be sure you have .NET installed for that)
-
I've tested it too and no result. I've no error message. u have the SAPI SDK 5.1, a headset microphone. I trained the software with my mouth. And when using the -voicer flag, and say into the mic "Alpha two destroy my target" or "Beta cover me" there are nothing done.
I use the latest CVS build.
-
Is this in the CVS? RT said he had problems connecting to cvs. Try the file in the first post
-
@metal : Did you train the thing to respond to your voice using the Speech panel?
-
Okay, I managed to make it work, and it really whips the llama's ass, as they say... The SDK installation solved the problem. Thanks to knn.
It's actually quite good, only thing is that you have to divide the command into two parts: Alpha wing *break* Destroy my target. During the break it opens the comm window, and if it doesn't open that, the latter part of the command does nothing. But it's still cool, and managing different tasks to different wings is quite a bit faster and more accurate.
I just might try the Shoot with IL-2 FB...
-
That could be solved by making a new command 'Alpha wing destroy my target' instead of 'Alpha wing' and 'destroy my target'. That would also mean you would have to add every other command plus every other wing name... lots of commands. Here's where having a built in engine comes in handy: instead of checking for simple matches, it could record the entire sentence, then check the first word, if it matches a wing name, continue and try to recognize the command. Wing could be optional, so 'Alpha, destroy my target!' would work as well. Since I don't know a thing about MS SAPI programing, I don't know how difficult this is to implement, but if it's possible, it would give voice enabled FS2 a huge advantage over FS2 and GameCommander/Shoot
-
@metal : Did you train the thing to respond to your voice using the Speech panel?
Yep sir ! I train the thing ^^. I will try to give a break between Alpha Wing and Destroy my target.
-
You only have to say
"alpha, destroy my target"
-
Hey RandomTiger
I just found this build and I have to say this is the coolest new feature I've seen in awhile! :yes: It makes really heated battles so much easier to manage. I'm wondering how tough it would be to add even more commands. Any chance of putting in things like shield managment, targeting, match speed, etc? I know there's the shoot! option, but this seems so much more elegant.
-
Got a problem. I'm running XP Professional, and I've got no Voice recognition option. When I select "speech" in the "control panel", my only options are Text-to-speech. It seems that I need Office XP in order to get it. So much for that idea.
-
Installing the speech SDK should work.
-
Installing the speech SDK should work.
Do you mean THIS (http://download.microsoft.com/download/speechSDK/SDK/5.1/WXP/EN-US/speechsdk51MSM.exe) or THIS (http://download.microsoft.com/download/speechSDK/SDK/5.1/WXP/EN-US/speechsdk51.exe)? 'Cause I already downloaded and installed the first of those, and all it had was some compressed files.
[Edit]
OK, I assumed that being the larger file, the first one would have everything the second one did. Installing the second option has solved my problem. Now all I have to do is find a patch cable for my mic.
-
tested this and its now my favourite build <3 but for some reason thruster trails don't seem to work with it. will try a few other things... (more <3)
-
double post i know, but i really don't want to see this feature buried. any chance of an 02/28 or 03/01 build with -voicer functionality?
-
Um, all recent builds have this. I haven't downloaded the 03/01/06 build, but the 02/28/06 build has it. The checkbox is in the experimental category of the features section.
-
gives me an "unable to load resource" error and then "unable to init voice rec" error on start (then runs fine) though. i will investigate.
*edit*
works fine with original voicer build, but not with 02/28 or 03/01, go figure. anyone else experienced this?
-
Confirmed. The last CVS build to function properly with -voicer was 2006/02/22. All since give the same error message:
"Error: Failed to load resource"
followed by:
"Error: Failed to init voice rec"
Anyone able to provide some insight on this?
-
*bump*
3.6.9 RC3 also gives the same two messages.
Original voice recognition build works - on behalf of voice recognition, that is. Other things are not as good as with newer builds, OGL env mapping is not there, and it's not as stable.
Furthermore, Taylor's 20060513 -build doesn't give the error reports, but the recognition doesn't seem to be working in-game.
-
Recognition is broken and no one except Random Tiger ever understood how it worked in the first place so it's likely to stay broken.
-
Recognition is broken and no one except Random Tiger ever understood how it worked in the first place so it's likely to stay broken.
So will someone take it out atleast?
-
Recognition is broken and no one except Random Tiger ever understood how it worked in the first place so it's likely to stay broken.
So will someone take it out atleast?
It's already a build option and disabled by default. But if RT doesn't come back and fix it soon then we'll probably just rip the code out so that we don't have to deal with it.
-
Decided to have a play with source for this and managed to add a few more commands.
<RULE ID="VID_targeting" TOPLEVEL="ACTIVE">
<L PROPID="VID_targeting">
<P VAL="0">Target next</P>
<P VAL="1">Target previous</P>
<P VAL="2">Target hostile</P>
<P VAL="3">Target previous hostile</P>
<P VAL="4">Auto target</P>
<P VAL="5">Target Friendly</P>
<P VAL="6">Target previous friendly</P>
<P VAL="7">Target sight</P>
<P VAL="8">Targets attacker</P>
<P VAL="9">Target sender</P>
<P VAL="10">Stop targeting</P>
<P VAL="11">Target sight subsystem</P>
<P VAL="12">Target next subsystem</P>
<P VAL="13">Target previous subsystem</P>
<P VAL="14">Stop targeting subsystems</P>
<P VAL="88">Target next escort</P>
<P VAL="89">Target closest repair ship</P>
<P VAL="90">Target next uninspected</P>
<P VAL="91">Target previous uninspected</P>
<P VAL="92">Target newest</P>
<P VAL="93">Target next turret</P>
<P VAL="94">Target previous turrent</P>
<P VAL="95">Target next bomb</P>
<P VAL="96">Target previous bomb</P>
</L>
</RULE>
<RULE ID="VID_speed" TOPLEVEL="ACTIVE">
<L PROPID="VID_speed">
<P VAL="15">Match speed</P>
<P VAL="16">Auto match</P>
</L>
</RULE>
<RULE ID="VID_power" TOPLEVEL="ACTIVE">
<L PROPID="VID_power">
<P VAL="69">Increase weapon</P>
<P VAL="70">Decreas weapon</P>
<P VAL="71">Increase shield</P>
<P VAL="72">Decreas shield</P>
<P VAL="73">Increase engine</P>
<P VAL="74">Decrease engine</P>
<P VAL="75">Power balance</P>
<P VAL="105">Max engines</P>
<P VAL="103">Max shields</P>
<P VAL="101">Max weapons</P>
</L>
</RULE>
Compiled verson in from a cvs grab on 28 Oct, http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip (http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip).
Not entirly sure if all the targeting works, but the power management is fine.
The modified files are here http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip (http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip)
-
Had a bit of playing with code for this. Managed to remove the need to pause in most phrases, so you can say "Alpha 1 Destroy my target" rather than "Alpha 1" pause "Destroy my target".
The phrase file is now also load at runtime so if a phrase isn't the way you want it, it can be edited.
Hopefully it will also work for other than english language settings.|
http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip (http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip) (4.36Mb) - modified version of the 3.6.9 branch code - grabbed from cvs on the 6th Nov -
only contains standard exes and a copy of the phrases.xml to place in the data folder.
http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip (http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip) (5K) is the modified version of the voicer.cpp file.
I've also attached a copy of the phrases.xml to this post. Enjoy.
I find using this better than extra programs mention previously, no pauses required in the phrases.
[attachment deleted by admin]
-
Just exactly how accurate is this now? I mean, half the time when I use voice-recognition with Game Commander 2, I always seem to get a few hundred commands mixed up.
-
Probably about the same sort of accuarcy initialy. The Microsoft stuff can be trained up further from standard if required.
-
Cool. :) Shall I commit this to CVS? (It would be in the unstable branch and wouldn't appear in 3.6.9, but it would be in future releases.)
-
If you want to stick in there be my guest. I find it handy, if others do as well thats even better.
-
Probably about the same sort of accuarcy initialy. The Microsoft stuff can be trained up further from standard if required.
So, without voice training, commands can still be mixed up. Got it.
-
K, committed.
-
awesome. Wish it was in 3.6.9, but there's always the CVS builds :pimp:
-
MadProf: what version build is your build based off of?
-
The version linked to is a modified 2.6.9 from a few weeks ago.
-
link down
edit: silly me, it was an outdated one.
-
The version linked to is a modified 2.6.9 from a few weeks ago.
WOW! Talk about your retro build!
-
Any chance that voice recognition (and MadProf's fixes and extensions) will ever hit the official 3.6.x branch?
And, before I make myself look the fool: a question.
What, exactly, is the difference between all the different branches in the code? CVS, official 3.6.x, HEAD.... What is the difference bettween all three?
-
Any chance that voice recognition (and MadProf's fixes and extensions) will ever hit the official 3.6.x branch?
It should be in 3.6.9 final. If it's not, let us know and we'll rebuild it.
And, before I make myself look the fool: a question.
What, exactly, is the difference between all the different branches in the code? CVS, official 3.6.x, HEAD.... What is the difference bettween all three?
CVS is the program used to manage the code so that we can keep track of all the changes we make to it.
HEAD is the main version of the code, where all the changes (both features and bugfixes) go.
The 3.6.9 branch is equal to HEAD except that it has none of the features added since June 2006. We created this branch so that we could focus on bugfixing while making sure not to introduce any new bugs.
-
Makes sense, but what are he CVS builds, then? What kind of features are missing from the 3.6.9 build that are in HEAD?
-
Makes sense, but what are he CVS builds, then?
CVS builds are occasional builds based on whatever code is current. They can be from either HEAD or the 3.6.9 branch.
What kind of features are missing from the 3.6.9 build that are in HEAD?
As I said, everything since June 2006. Two examples are karajorma's custom loadout code and my custom command persona code.
-
Or for the non FREDders there are side multipart turrets and WMCs improved scripting system. :)
Thing is though that unless you're making your own campaign HEAD is not worth the hassle of using at the moment. 3.6.9 is more stable. HEAD is for early adopters and developers. 3.6.9 is for everyone else.
While I'm at it we really should change RedMenace's topic to HEAD CVS builds to avoid confusion :)
-
YOU ARE SUPPOSED TO AVOID THE BEAM CANNON! DAMMIT!!!
-
Decided to have a play with source for this and managed to add a few more commands.
<RULE ID="VID_targeting" TOPLEVEL="ACTIVE">
<L PROPID="VID_targeting">
<P VAL="0">Target next</P>
<P VAL="1">Target previous</P>
<P VAL="2">Target hostile</P>
<P VAL="3">Target previous hostile</P>
<P VAL="4">Auto target</P>
<P VAL="5">Target Friendly</P>
<P VAL="6">Target previous friendly</P>
<P VAL="7">Target sight</P>
<P VAL="8">Targets attacker</P>
<P VAL="9">Target sender</P>
<P VAL="10">Stop targeting</P>
<P VAL="11">Target sight subsystem</P>
<P VAL="12">Target next subsystem</P>
<P VAL="13">Target previous subsystem</P>
<P VAL="14">Stop targeting subsystems</P>
<P VAL="88">Target next escort</P>
<P VAL="89">Target closest repair ship</P>
<P VAL="90">Target next uninspected</P>
<P VAL="91">Target previous uninspected</P>
<P VAL="92">Target newest</P>
<P VAL="93">Target next turret</P>
<P VAL="94">Target previous turrent</P>
<P VAL="95">Target next bomb</P>
<P VAL="96">Target previous bomb</P>
</L>
</RULE>
<RULE ID="VID_speed" TOPLEVEL="ACTIVE">
<L PROPID="VID_speed">
<P VAL="15">Match speed</P>
<P VAL="16">Auto match</P>
</L>
</RULE>
<RULE ID="VID_power" TOPLEVEL="ACTIVE">
<L PROPID="VID_power">
<P VAL="69">Increase weapon</P>
<P VAL="70">Decreas weapon</P>
<P VAL="71">Increase shield</P>
<P VAL="72">Decreas shield</P>
<P VAL="73">Increase engine</P>
<P VAL="74">Decrease engine</P>
<P VAL="75">Power balance</P>
<P VAL="105">Max engines</P>
<P VAL="103">Max shields</P>
<P VAL="101">Max weapons</P>
</L>
</RULE>
Compiled verson in from a cvs grab on 28 Oct, http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip (http://www.intelwatch.demon.co.uk/fs2open/fs2open_voicer.zip).
Not entirly sure if all the targeting works, but the power management is fine.
The modified files are here http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip (http://www.intelwatch.demon.co.uk/fs2open/modifiedfiles.zip)
Where exactly would I place the contents modifiedfiles.zip. or even the voicer.exe? I have BTRL and FS open 3.6.9 with the voice rec box checked in the launcher it works but I want to add the targeting and power management with the modifiedfiles.zip?
-
Sorry for necro, but I just realized that this was this guy's first and only post here, and it never got answered or welcomed. So, first thing's first:
:welcome-red:
Welcome to the HLPBB! Sorry it's 4 months late.
In case anyone does run across this (hell, it was still on the first page of topics at least), those files are source code, and would have to be compiled in with the rest of the SCP source code. You can't just drop them into your regular install folder and have them work.
-
is this feature still working?
-
Not in nightlies. 3.6.14 RC1 will have it enabled.