Recommendations : Voice Control

Nice thing about girder is, you basically can control anything windows can do. Therefore, the speech plugin just attachs a speech input to any macro you create in Girder.

Girder 4.0 is around $40 retail, but if you're crafty (smile) you can still find version 3.29b which is freeware.

The speech plugin can still be found on their site.
http://www.promixis.com/downloads.php?mode=list

This speech plugin only works for top level grammar, so dictation is way out. But for voice commands, so far, I haven't found a limit.

As soon as I finalize and test my command structure, I'll post the GML and XML.

downloaded the girder plugin and i havent the FIRST clue how to set up commands. Where did you learn to set your xml/gml up?

LOL...Trial and error. The README doesn't help too much.

Here's the deal; the speech plugin pulls from the XML file. If you downloaded the plugin there's an example XML file (that must be put in the PLUGINS directory to work).

You need to add to the XML for any voice tag you want to create. Use the two examples to guide you if you're not familiar with XML.

Here's the XML file as it comes with the plugin I've added a third in bold...

<GRAMMAR LANGID="409"> <!-- american english grammar -->

<DEFINE>
<ID NAME="VID_Test1" VAL="1"/>
<ID NAME="VID_Test2" VAL="2"/>
<ID NAME="Launch_Mappoint" VAL="3"/>
</DEFINE>

<RULE ID="VID_Test1" TOPLEVEL="ACTIVE">
<P>computer</P>
</RULE>

<RULE ID="VID_Test2" TOPLEVEL="ACTIVE">
<P>quiet</P>
</RULE>

<RULE ID="Launch_Mappoint" TOPLEVEL="ACTIVE">
<P>Launch Map Point</P>
</RULE>
</GRAMMAR>

Basically, what this does is creates the voice tag to compare against SAPI. The top section (DEFINE) just defines the identifier, the second section (RULE) lists what word it's to look for. Then create a command in Girder for anything you want, select the speech plugin, click LEARN, and say COMPUTER. It should recognize it as identifier 1 and initialize whatever command you attached to it.

Some recommendations I can make out of the gate:
1) TRAIN YOUR PROFILE!!! Even if it's only the basic training in windows you have to give it a basic grammar to work with
2) Keep your commands short, but not too simple. You want the commands to be easy to remember but not simple enough that they could be picked up mistakenly in normal conversation
3) Sometimes it's better to spell things out phoenetically in the XML file. SAPI works off of each individual letter and may not pick up on your accents. Spell it like it sounds!

Sweet!

I use girder extensively on my HTPC and I absolutely love it! I'm a huge fanboy of girder because you can do just about anything with it. I hadn't even thought to look for a speech plug-in for it! That's the best news for my carpc I've heard in months.

Have you tried to use pre-existing gml files for various media players with the corresponding front-ends (i.e. WMP and FP, Winamp and RR)? Just curious, as that would be easier to modify a pre-existing one than to start from scratch.

Great find Wolfebaine!

And I'm definitely interested in your gml and xml files as you go.

So how can you trigger girder to listen for commands? hotkey? spoken command, a la Navivoice? or is it constantly listening?

Sorry for the frantic response, I'm just excited about the possibilities! Did I mention I like Girder... ?

ok i think i understand.. isnt too different from navivoice. But what i'm confused about is where do i put the command? I dont see a command in the example you posted.

Quote:

---

Originally Posted by Sonicxtacy02

ok i think i understand.. isnt too different from navivoice. But what i'm confused about is where do i put the command? I dont see a command in the example you posted.

---

What he listed was just how to get girder to recognize the voice command "launch map point". Before you do anything you should go through the steps that Wolfebaine listed above. Once that's done and you've added the wolfebaine's command listed previously, go into girder, hit Edit at the top, choose New Command and an object called "New" will be created. There are several ways to do this, but the one I'll use is probably the easiest to do while just starting out with girder.

With the "New" command highlighted, on the right-hand pane, select the O.S. tab and choose "Execute" from the top drop-down box. Now click "browse" and find the executable you want the voice command to run (in this case, Map Point). Once that's done, hit "Apply".

Now you're ready to set the event that will trigger the command (in this case, a voice command but girder supports ir commands, keyboard input, other commands can trigger a group of commands... the list goes on) Click "Learn Event" at the top of the right pane and say "launch map point". If SAPI and the xml file is setup correctly, it should display the value that the command was given in the xml (3 in this example). Say it again and it should launch the program. I don't use MapPoint so I don't know how you may be using it. It may be a little more complicated if the app is embedded, but if there's a hotkey associated with it just use the Keyboard tab instead of the O.S. tab as described above and select the appropriate key combo.

Hopefully that will get you going in the right direction.

Quote:

---

Originally Posted by wolfebaine

LOL...Trial and error. The README doesn't help too much.

---

That's too bad cause it's a well documented format actually.

Technical description: This file is defining the grammar (ie Context Free Grammar, or CFG) that the speech recognition engine is using. It is using two schemas (Note: right click to save, or use View Source), the
SAPI schema that is for the speech engine to control pitch, volume, speed, voices, all kinds of cool things. the grammar rules, etc, and for the grammar schema that defines the CFG. MSDN has some Great documentation for it. Note this requires not just XML & XML schema knowledge but how context free grammars work, something I first learned of in CS Theory :nutz: But it's fun stuff, and not too bad IMO.

Somewhat Simpiler answer: XML Text-to-Speech tutorial can walk you through how to change pitch, speed, define pheonomes, etc.

@Sonicxtacy02: Yes they're both macro interfaces really (ie they basically "convert" your voice commands to keyboard/mouse commands).

Looks like someone has another(newer) version of the speech plug-in for girder.

http://www.promixis.com/phpBB2/viewt...hlight=timeout

Looks simpler to use.

Nevermind - sorry it's for another Promixis product, NetRemote

Lord - drawback is Girderspeech is that it's constantly listening for commands. So mistaken recognition is probable. This is where I'm playing around and having some success.

Here's a tatste of what I'm working on : I've made a group in girder called "Voice Commands" which, by default, is inactive. Above that, I've created a command called "Hal". All this command does is activates the "Voice Commands" group. Every command within this group is a multicommand that contains the eventstring(speech), the command that executes, and finally a command that disables the "Voice Command" Group (wow...i used the word command too many times).

End result? All voice commands only respond when the word "Hal" is spoken first.
"Hal <activates Voice Command group> Play All Music <executes then disables the voice command group>"

Now, next step is for me is to figure out how to have the Hal command timeout after a few seconds if no command is said after that (meaning, the Voice Command group goes inactive again).

Attached is the XML/GML for just the stuff I've mentioned above. Have fun!!

Wolfebaine,
I believe this is what you're looking for. I just finished some research on that same thing (except I wanted a button trigger instead of a voice command to start the "listening"). To change the voice trigger just Learn Event for the "Speech Trigger" multigroup (currently set to run with the example "computer" command)

It's set to enable the example group after 4 seconds then disable again, but you can modify the time in the settings of the "Delay Event" command.

Enjoy!