By Sy Mukherjee
April 24, 2019

Hello and happy hump day, readers.

University of California, San Francisco researchers have done something truly remarkable. In their words: “We designed a neural decoder that explicitly leverages kinematic and sound representations encoded in human cortical activity to synthesize audible speech.”

Ok, let’s simplify that a bit. The scientists were able to translate brain waves into intelligible speech, including full sentences that even mimicked study participants’ natural speaking cadences, according to preliminary research published in the journal Nature.

The technology involved placing electrodes on the surface of severe epilepsy patients’ brains in order to record the brain activity associated with speech. A computer was then used to create an algorithm that could then decode those signals—and the result was a virtual speech generator which, theoretically, could one day fuel technology to restore speech to stroke, ALS, and other neurodegenerative disorder patients.

In fact, that’s the next step in the project—to test out the technology on people who have already lost their ability to speak. The NIH-funded project is, so far, a proof of concept. But it’s an extraordinarily tantalizing concept.

Read on for the day’s news.

Sy Mukherjee


You May Like