Subvocal speech recognition

If any of you have used speed recognition software you know how shakey it can be. You also know how useful it can be when it is working well. NASA and researchers at Carnegie Mellon are working on a similar technology but designed to read subvocalizations. According to this article it is drastically worse than present speech recognition. If they get this up and running well (which no doubt they will) we may be seeing a new peripheral for our cellphones. Or perhaps IPOD's since word on the street is that they may be hitting the cell phone market soon enough.