API link for Creating Voice Attack Plugins

[Chris McIntosh]

Hey gang I was very much interested in creating some voice attack plugins, does anyone have a link to any resources about doing this.  Also, what languages would be proper (ie C#, VB.net, etc).

thanks

[Luca D]

Any .NET language should do; as for the development environment, the free, Express edition of Visual Studio is enough

The VA help file has all the details about the public interface you need to expose; as for some examples, have a look here:
https://github.com/Antaniserse/VAExtensions

It's still undocumented, but as a reference, you can take the same sample profiles and help document included in the old version, here:
https://dl.dropboxusercontent.com/u/90846278/VATextReader.zip

If anything is unclear in the code, feel free to ask

[Chris McIntosh]

I appreciate the feedback Ill take a look at things now.  Additionally, the one thing that I think seems to be lacking atm from my experience is the option for what I will call Vocal Arguments.

Meaning you have a command: Search for {term} on google

In my mind what would happen is Voice attack would take the word or immediate phrase after Search For, then pass that as text available in the {term} variable.  This would allow for a lot more versatile experience when building out profiles.

[Luca D]

That's just the kind of thing you can do relatively easily with a plugin.

Since VA now supports wildcard, you could define a command that responds to "Search for *", make that execute a plugin call, and in your code parse the whole command (which you can pass to the plugin with the already existing tokens) to extract the desired {term}, maybe even the "on google" part, if, say, you want to use different search engines for the same command), and finally launch the proper query

Keep in mind however that non-dictionary/not know before hand words (like anything you might say as {term} and even the word "google" itself) are hard to catch properly by the engine, so while the command while fire because it will recognize correctly the first, fixed part, you are gonna have, in many cases, a lot of garbled stuff in the remaining string.

[Chris McIntosh]

Out of curiosity is there any plans or thought regarding using something like Sphinx instead of Microsoft's engine, is the performance/dictionary the same?

I had messed around with Sphinx a bit with a Java app that I was toying with and it seemed to recognize things pretty well (actually so does VA).

[Gary]

Sorry, Chris.  There are no plans for Sphinx as of right now.  Actually, I'm more on the 'nothing extra to install' side of the fence.  Also, there's a LOT of code in VA that deals with quirks surrounding the speech engine, and I would have to write code to handle any new speech engine's oddities to keep VA working right.  The closest thing would be to implement the Microsoft Speech Platform with its various additional speech languages first (even though it does not have dictation).

Does the Sphinx engine handle dictation better than the built-in Microsoft speech engine? The dictation is the only thing I'm not very impressed with in Windows' speech engine (hoping Windows 10 has a better engine).

Gary