Yeah, I've got SAPI (Microsoft's Speech API) working. I'd prefer the Narrator though - although it uses the same voice, its something totally different (it can read texts word by word or even letter by letter, and much more. It even can control where the 'focus' is with gestures and/or keys)
But since I found nothing (only how to create programs using labels and etc), I'm going for now with the SAPI indeed. But then, something else jumped on me: I thought to read the configs from Narrator (pitch and rate of the voice) and apply it on SAPI to mimic the narrator voice.
I can read it just fine in the registry, but applying the settings on SAPI doesn't quite behave the same way. I've made a post on StackOverflow about it, let's see if I get some cool answer there to solve this one
Sorry the final text spoken is in Brazilian Portuguese, but I think you guys can understand the gameplay. The screen goes darker by the second, and after a while you'll see nothing - still having to avoid the monsters just by using your ears (and feeling joystick vibrations if you're using one capable of)
Nope - that is just to help identify "objects that are accessible to Narrator" (which means buttons, labels, text boxes, etc)
Those "commands" mentioned in the article are just a bunch of key combos that activates special features, like highlighting the current Narrator focus, for example.
To be able to use the "Narrator" AFAIK, you need to use standard controls. Using OpenGL/Direct-X and images won't work, at all :/