Theoretically it is possible - the DBPro plugin DarkVoices uses the Microsoft Speech Engine SDK 5.1 (or elements of it) to generate lipsync data for an animated head model. Effectively, it uses voice recognition to convert a sound to a string of text.
Now, the challenge is first to find the speech engine (no mean feat, Microsoft pulled it from their site some time ago, but it is still out there I think). Second, you need to learn C++ so you can access the speech libraries. Thirdly, you need to produce a function library (a dll) that DarkBasic can call using "Load DLL". Finally, you need to take care of all those things like "What happens if the speech engine doesn't recognise the words I said properly?"
This is all possible, but I think you'd do better to stick to keyboard mouse and gamepad for input devices. Or you could beg TGC do make a plugin for it and upgrade to DBPro.
(Actually, don't knock that last idea. GreenEar refers to voice recognition on its website, but although I asked about it, it doesn't appear to be in the DBPro build.)
"I wish I was a spaceman, the fastest guy alive. I'd fly you round the universe, in Fireball XL5..."