New ‘mind reading’ system makes synthesized speech more human – CNET

For years researchers have tried to convert brain signals directly to speech, cutting out the muscles we actually use to speak. A new approach out of UCSF puts the jaw back in jawing.