A neural prosthetic device developed at Boston University has partially restored the speech of a mute human volunteer. The brain implant decodes neural signals from the brain's speech motor cortex and uses them to operate a formant synthesizer to reproduce speech. So far, researchers are concentrating on producing individual vowel sounds before progressing to words and sentences. The formant synthesis method is based on a mathematical acoustic model rather than by concatenation of sampled words or morphemes. The video above shows the subject learning to pronounce the vowel sounds "uh-oo". This research is being done by Professor Frank Guenther and his team. For more details on how it all works, see the team's 2007 Society for Neuroscience conference poster (PDF format) or the most recent paper from Frank Guenther, Neural mechanisms underlying auditory feedback control of speech (PDF format). via naturenews.