Boston University researchers are developing brain-reading computer software that is in the early stage of translating thoughts into speech, starting with vowels.

An implanted electrode picks up nerve signals related to movement of the mouth, lips, and jaw. These signals are sent wirelessly to a computer, where software analyzes them for patterns that most likely denote a particular sound, generating formant frequencies (the resonant frequencies of the vocal tract). These are converted into sounds via a speech synthesizer and played through a speaker for feedback and training.