‘Thought to text’ tech may soon let paralyzed patients speak again

‘Could scientists turn thoughts into speech? Hope for paralyzed patients as researchers develop implants to read the mind’s ‘inner voice’

  • Paralysis and neurological disease like ALS can destroy the neurons that control the physical act of speech 
  • This doesn’t mean they have lost language in their brains’ ‘inner voices’ 
  • Scientists at Columbia University and Northwell Health are implanting brains with electrodes to listen to the brain’s language 
  • Now, they’re starting to translate electrical impulses into simple words 

Scientists are getting closer and closer to being able to translate the thoughts of patients who have lost their speech into words. 

Technology has made leaps and bounds to bridge the rift that forms between mind and body when one (or both) are damaged by disease. 

Amputees can now be fit with futuristic mind-controlled body limbs; bionic eyes give sight to the sightless, cochlear implants allow the deaf to hear. 

Restoring speech, however, presents unique challenges and has mostly remained in the realm of science fiction and the (mostly) pseudoscience of telepathy. 

Ambitious scientists scattered across the world are on the brink of changing that with implants that act as ‘brain-computer interfaces’ that may soon be able to broadcast the voices inside a speechless patients’ heads.   

Researchers at Columbia University and Northwell Health are using electrodes on volunteers’ brains to learn the electrical impulse language of the brain so they can translate it and give speech back to paralyzed patients (file) 

  • Ex-Atlanta Falcons linebacker Tim Green announces he has ALS…

    ‘Playing ping-pong with people’s lives is inhumane’: ALS…

Share this article


Up until his death in March, British cosmologist Stephen Hawking was one of the most influential and famous thinkers, teachers and speakers of the last century. 

But for the last 30 years of his life, he had no voice. 

Computer assistance allowed the brilliant astrophysicist to go on to speak all over the world nonetheless. The voice of ‘Perfect Paul’ – an early synthesized speech tool with an American accent designed for telephone prompt responses – became Hawking’s signature tenor. 

Synthetic voice technology improved, but Hawking stuck with Paul’s slightly halting speech. He did, however, progress through various control mechanisms.  

Once his thumb could no longer choose letters, an innovative system that picked up on the subtle twitches of his cheeks allowed Hawking to go on communicating. 

But Perfect Paul never did read Hawking’s brilliant mind directly. ‘Speaking’ meant a letter-selection process that – even with the latest and greatest predictive text technologies – would always undoubtedly be slower than the great scientist’s natural speech and meditations on the history of time. 

ALS-stricken Stephen Hawking used a computer-assisted communication device that translated his thumb or facial movements into words – but it didn’t read his mind


Now, neuroscientists and engineers at Columbia University and Northwell Health in New York are mapping the brain’s private language so that they may soon be able to translate the voice inside our heads. 

Different regions of the brain communicate with one another through a combination of electrical impulses and chemical messengers.  

Listening and speech are centered in Broca’s area, which is responsible both for the brain’s voice and the ‘mind’s ear,’ as well. And Wernicke’s area controls our word choices. 

The collaborators at Columbia University and Northwell Health in New York hope that by listening in on the brain’s most intimate conversations with itself, they might learn how to translate its electrical language into words the outside world will understand via a brain computer interface, Stat News reports. 

This would be life-changing to patients with paralysis from injuries, ALS or locked-in syndrome.  

ALS often eventually attacks the bulbar neurons, which control the motor function or movements involved in the physical act of speaking.  

Loss of that function doesn’t mean that language is lost altogether – we just need to learn to understand the brain’s language. 

So Dr Nima Mesgarani, an electrical engineering professor is listening in on brain chatter using implanted electrode strips, according to Stat News. 

His electrodes record the electrical cracks and snaps in the brain, and sends them to a computer that then tries to make heads or tails of them. 

Just like any other language, Dr Mesgarani’s computer needs to keep learning more and more of the brain’s electrical vocabulary. 

So Dr Mesgarani has partnered with Dr Ashesh Mehta and his volunteers, epileptic patients who are allowing the scientists to implant their brains with the electrodes during surgeries to find the source of their seizures.  


The more patients they implant, the better they’ll know how to identify words. 

The first step is to learn what our brain’s electrical activity looks like when we say ‘yes’ or ‘no’ to ourselves. 

We are still a long way off from translating complex thoughts, however, and these will likely remain locked away in the mind for some time to come.

But neither they nor the handful of other scientists working on similar medical technologies are the only ones interested in brain-computer interfaces. 

Facebook, Elon Musk’s Neuralink, and other technology companies have launched similar research and development efforts – with less philanthropic purposes in mind. 

In 2017, Facebook announced its own goal of making a Face Brain, so to speak, that promises thought-to-text technology, so the Internet never has to miss out on a single thing you think. 

The Facebook team has insisted that thought-to-text is no longer science fiction, but Dr Mesgarani and Dr Mehta are taking a more measured approach. 

The five patients they recruited then had to painstakingly repeat numbers and stories until the doctors had recorded their brains’ electrical impulse patterns. 

These signals were then fed into a kind of medical synth, to see if they’d come out like language or gibberish. 

Dr Mesgarani and Dr Mehta got something in between. 

Most prior systems had only produced ‘words’ that were about half-and-half nonsense and comprehensible. The new brain translator was about 75 percent ‘intelligible.’ 

‘We have a good framework for producing accurate and intelligible reconstructed speech from brain activity,’ Dr Mesgarani told Stat. 

And that’s ‘a step toward the next generation of human-computer interaction systems … for patients suffering from paralysis and locked-in syndrome.’       

Source: Read Full Article