Recent feats of science give a whole new meaning to "thinking out loud."
UC San Francisco neuroscientists succeeded in using brain recordings to create synthetic speech, according to research published in Nature, an international science journal, on April 24, 2019. This technology could change lives for people who otherwise can't communicate due to neurological impairments.
Researchers Gopala K. Anumanchipalli, Josh Chartier and Dr. Edward F. Chang described in their abstract that it's challenging to decode speech from brain activity.
"Speaking requires very precise and rapid multidimensional control of vocal tract articulators," the abstract stated. "Here we designed a neural decoder that explicitly leverages kinematic and sound representations encoded in human cortical activity to synthesize audible speech."
So What Does That Mean?
Basically, these scientists created and used a brain-machine interface to generate synthetic speech that sounds natural from brain activity, as reported by Nicholas Weiler on UCSF's website. The machine utilized neural activity to control a virtual vocal tract, consisting of a computer-simulated lips, jaw, tongue and larynx.
"For the first time, this study demonstrates that we can generate entire spoken sentences based on an individual's brain activity," Dr. Chang said, according to Weller's reporting. "This is an exhilarating proof of principle that with technology that is already within reach, we should be able to build a device that is clinically viable in patients with speech loss."
How Did They Do It?
For their study, Chang and his team used data from five patients whose brains were being monitored for epileptic seizures, as reported by National Geographic. Each participant already had arrays of electrodes, each about the size of a stamp, placed on the surface of their brain. The participants read off hundreds of sentences as the electrodes monitored brain activity and the brain-machine interface translated this activity into speech.
Christian Herff, a Maastricht University postdoctoral researcher who studies such speech methods, called this study a "very, very elegant approach."
Why Does It Matter?
Neurological damage can result in an irreversible loss of the ability to speak, according to UCSF. Such damage can come from traumatic brain injuries, strokes or neurodegenerative diseases, such as Parkinson's. People who suffer from speech disabilities often cope with devices that use eye and facial muscle movements to spell out their thoughts, letter-by-letter. However, this mode of communication is tedious and inaccurate, and doesn't resemble natural speech.
Chang's work may change that. Where current communication devices allow speech at about 10 words per minute (or fewer), his team's research allows communication technology to work closer to 100 to 150 words per minute – the rate at which most people naturally speak.
What's Coming Next?
Scientists still have a long way to go to make this technology as accurate as possible, and it's unlikely to help people with severe damage to the speech centers of the brain. More viable users simply lack control over their speech muscles.
Melanie Fried-Oken, speech-language pathologist at Oregon Health & Science University, told National Geographic that while this research raises some ethical questions regarding identity and privacy of thought, it also holds promise.
"Wouldn't it be great to be able to give this to a 3-year-old who can now interact with the environment, who hasn't been able to do it yet?" Fried-Oken told National Geographic. "Just like we're giving cochlear implants to [deaf] infants – the same. There's just such potential here, but there's so many neuroethical issues."