Scientists create speech using brain recordings

SAN FRANCISCO, CALIFORNIA — New research from the University of California San Francisco found that brain recordings can be converted into speech.

According to a paper published in the journal Nature, neuroscientists from UC San Francisco have developed a technology that can convert brain recordings into synthetic speech using a computer simulation model with lips, jaw, tongue, and larynx.

The team recorded brain signals of subjects that already had electrode arrays implanted in their brains for previous medical procedures. Subjects would read sentences out loud while the researchers captured the signals through the electrodes.

They then analyzed the audio and observed which muscles were activated at different points during speech.

Using this information, they developed a simulation that would transform brain activity into muscle movements in a virtual vocal tract and in turn, turn the movements into a synthetic voice.