No announcement yet.

Voice Recognition Software for games!

  • Filter
  • Time
  • Show
Clear All
new posts

  • Voice Recognition Software for games!

    I had originally posted this on (the post is credited to the person that started the news item, i wrote almost everything but the first few sentences) but I couldn't let this nugget of info pass without letting you guys know about it, because I feel it could be used effectively in many of the games we play, as sort of a 'helping hand', although I haven't been able to extensively test the program completely yet. Hold onto your butts, cause this is going to be long...

    Shoot is a speech recognition program that translates your voice commands into arbitrary keystrokes. Perfect for commanding your AI buddies in your favorite flight simulator or FPS! Forget about having to memorize all those keyboard shortcuts, now you can just say what you want to happen. Although it was created with games in mind, it should be easily adaptable to practically any program that uses shortcuts or other keyboard commands by creating an appropriate command profile, thereby allowing you to control editing functions in Photoshop, Word, and others by using your voice.

    First things first. Shoot requires a SAPI compatible voice recognition engine. Those of us with Office XP and it's associated voice recognition software should be able to get it working right off the bat. If not, I suggest downloading Microsoft's Speech Recognition SDK (If someone can point me to a more up-to-date one please tell me?) which i have tested and found to be compatible. It also requires .NET Framework 1.1 and Direct X 8.1 (a formality if you play most modern games).
    *SPECIAL NOTE* If you have .NET Framework 2.0, you will need to download a special fix to get it to be able to load profiles. The fix can be found here. Simply replace the existing shoot.exe.config file with this one. Alternatively, the associated Profile Editor contains the same fix. *END SPECIAL NOTE*
    After installing the SDK (if needed) and Shoot, I recommend training the voice recognition engine.

    This consists of taking 5-10 minutes reading phrases into the microphone to adjust the engine's sensitivity to your voice parameters (the SDK requires you to do it once it's installed). Continued training supposedly makes the engine recognize your voice better and better, I've found most responses are recognized after 2 training sessions.

    Now onto the fun stuff. The program's profiles are written in a special XML format, somthing I'l admit to not being to keen on. There are basic instructions on the documentation page, but fear not if you don't know how to code for it, there is a solution!

    There is a beta version of a Profile Editor available to download. Please note that it requires .NET Framework 2.0 to work, and as such, it requires the aformentioned fix if you need to run it (it conveniently also includes a patched file) The readme isn't quite so comprehensive, so I'l add in some of the things i've discovered and try to break it down for you (although, read the documentation anyway. Always read documentation..)

    First off, you'l need to create a new profile from the file menu. The two boxes marked Delay and Pause control a global variable that determines (in miliseconds) the delay between pressing and releasing a key and the pause (wait) between keypresses. You don't necessarily have to add anything here, because you can do that for individual keys if necessary.

    To create a new command, press Add under the command grid. now, type in a description of the command (after clicking in the highlighted box.) Now, in the Phrase section, type what YOU want to say to be registered by the microphone to trigger the command (for example, to trigger the communication command in Planetside, "I'm repairing the spawntubes" which required 4 seperate keypresses, I simply say into the microphone "repairing spawn tubes").

    To code the keypresses for the command, hit the Add button on the Action side of the menu. The dropdown box contains 4 selections:

    Key: One keypress. This can be any key on your keyboard (To add special keys such as the Function keys, printscreen, etc, highlight the next box and press F1 to get an illustration of a keyboard, whence you click on the keys with the mouse.)

    Press: Use this (Coupled with the Delay button) to hold a key down while pressing additional characters.

    Sequence: Use this for a string of keypresses together. Be sure to not add a space between the keys otherwise it will insert a space in the sequence. I couldn't figure out why i was jumping everytime I requested a medic until i looked at the profile and noticed I had left a space in there.
    Release: To stop holding a key held by Press.

    Now, add your keys to the actions. You can continue hitting Add to add more keypresses, etc for the command.

    Towards the bottom:

    Initial PTT State: This tells Shoot what the listening state is when you load the profile: whether it's listening by default, or off.

    Hold Action: This enables, with a held key, to either flip recognition off, on, or turn it off when it's normally on, etc.

    Sampling Rate: This adjusts the frequency of the samples Shoot and the voice recognition software takes of your voice through the microphone, and is NEEDED. Note that lower values are less taxing on your computer, but produce lower quality results and misunderstood commands, and may interfere with the sampling rate in the game or possibly teamspeak.
    The Toggle and Hold columns allow you a button or key to enable/disable the formentioned Push To Talk functionality.

    The sounds columns below these allows you to set up soundsamples letting you know when recognition is enabled/disabled, or if a command is understood or not.

    Remember to save your profile often, seeing as it's a beta, and theres always a possibility of a crash.

    Now, open up Shoot and load your newly created profile. In the bottom right corner is a microphone, this tells you if voice recognition is currently on or off. Go ahead and try out a few of your phrases, it will tell you in the bottom left if it's been recognized or not, or if you are too loud/soft. Enabling Send Keys will send your keypresses to whatever program you have open. This is all that's needed to add the voice functionality to the game or program you have open.

    There are some basic profiles included in Shoot, unfortunately some of them are for older versions of games so may not work. The Shoot website has more for other games, and
    has more, but I'd try to create my own or modify some to fit your own style.
    Happy Fragging! Perhaps we if this takes off we could make a profile database?

    <04:11:24> *** You are now talking in channel: "TFP - Task Force Proteus"
    <04:16:25> "|TG-XV| Tralic": this channel is so gay
    DICE needs to make a comical boxing glove attached to a spring punch the player in the face 40% of the time they get into a helicopter or jet.

  • #2
    Re: Voice Recognition Software for games!

    This would have really been nice back when I was playing Everquest. Neat program, will check it out.


    • #3
      Re: Voice Recognition Software for games!

      Great tool, might I say. I've recently stumbled upon this too and, despite not being a native English speaker, found it amazingly responsive after just two training sessions.
      I'm using it for Operation Flashpoint and I've adapted the profile to my own preferences (providing alternative calls such as 'all' used along with 'squad', 'cease fire' as 'hold fire' etc. and missing shortcuts like 'watch north' etc)
      The most dangerous thing on a battlefield is an officer with a map.


      • #4
        Re: Voice Recognition Software for games!

        I have the Microsoft GameVoice controller and software. Aside from the fact that TeamSpeak directly supports the GameVoice controller, the voice activation/control functionality of the software is quite nice. I've been using it for years. I tap my Command button and speak a command and the software does the rest!

        Diplomacy is the art of saying "good doggie" while looking for a bigger stick.


        • #5
          Re: Voice Recognition Software for games!

          I love this thread. Thanks for the information guys, what have I been missing?
          ~~ Veritas simplex oratio est ~~
          No matter how far a wizard goes, he will always come back for his hat. --T. Pratchett

          <---- You know you're getting old when you rely on your forum meta-data to remind you how old you are.




          TeamSpeak 3 Server


          Twitter Feed