what are you guys talking about? Naturally Speaking is up to 99 % accurate, and it’s always at least 95 % accurate for me and I don’t use it much. Voice recognition is here today if you want it. the only reason it’s not mainstream is because I’d rather click than say a link, and I’d rather type than say a word. (less effort)
They key is being able to trigger an event in blender that is transmitted from outside of blender. For example, the text is being inputed to windows notepad, and python is reading it. It can definately be done. Just depends on how interactive you need.
for example, write a very simple app in visual basic that is just a text box and make it save to a file whenever a word is entered. Then in blender, you can have a python script tied to a timer that every 3 seconds reads the text file and if it finds “left” in the text file, then spin the cube left, etc etc etc. The ONLY issue to worry about really is having that little app be in the background but still be the active window that Naturally speaking is putting text into.
If you don’t have access to Naturally speaking, I suppose you might be able to use the winXP built in recognition, but It’s not very good.
give a little more detail on what you’re trying to accomplish
by the way, the text to a file thing is a worst case scenario, hopefully, we can save a variable in memory in our little app that python in blender can access. but I haven’t tried it yet