Technology turns brain signals into words, helping polio patients communicate

This technology is expected to provide a voice for people with diseases like Parkinson's.

In the future, the technology to connect the brain with machines will help to return speech to those who have lost their voice by polio or by other factors such as throat cancer, muscle sclerosis on one side (ALS , also known as motor neuron disease) and Parkinson's.

Convert signals from the brain into pronunciation motion.(Source: The Guardian).

Some people - for biological reasons or circumstances - lose their ability to speak, for example, scientist Stephen Hawking has to say through a computer after suffering from ALS. At the time, Hawking could only use cheek movement to control a computer pointer that allowed him to say 10 words per minute.

But science and technology is now more developed. In an effort to design a method of directly converting brain signals into speech, scientists have created an artificial intelligence engine that can speak for people at natural speeds, about 150 words per minute. .

Edward Chang, professor of neurosurgery at the University of California San Francisco (UCSF) and author of the project, said: 'This is the first time we can create a complete credit-based statement. signal from the brain. This is clear evidence for the current technology, we can completely create devices that can help clinically lose patients' voice.

Speech synthesis devices , like what Stephen Hawking used, include spelling words one by one by tracking eye or facial movements. They allow users to speak about eight words per minute compared to an average of 100 to 150 words if spoken naturally.

Steps to convert brain signals into synthesized speech.(Source: Nature).

Kate Watkins, a professor of neurology at Oxford University, described the work as a "big step". "This plays an important role in providing people who are unable to communicate a device that can help them overcome this problem," Kate Watkins said.

Earlier efforts in converting brain signals to speech focused primarily on finding the presence of speech in the brain, and had achieved some success.

Chang and his colleagues tried a different direction. They aimed at areas of the brain that sent the instructions needed to determine the location of a series of sequential movements of the tongue, lips, jaw and throat during speaking.

Gopala Anumanchipalli, a speech scientist at USCF and the first author of this project, said: 'We think that if those brain regions encode signals into motion instead of speech, we So try doing the same thing in coding those signals. "

The team recruited five volunteers who are going to undergo neurological surgery for epilepsy. In the preparation for surgery, doctors temporarily implant electrodes in the patient's brain to map out the location of the sources of complications. While the electrodes are being implanted, volunteers are required to read hundreds of sentences over and over; while scientists will record the activity of the brain region responsible for making speech.

Dr. Chang and his colleagues combined recordings with data from previous sensors and experiments, to determine what signals control the movement of tongue, lips, jaw and larynx to help create sound.

The team trained a deep learning algorithm and included all the collected data. Thanks to the power of artificial intelligence, the algorithm has the ability to decode brain signals that convert them into estimated movements of sound-producing parts.

Sensors are attached to the patient's brain.

The machine that works on this algorithm has been able to collect brain signals to create artificial voices. Scientists recruited 17,000 other volunteers to listen to these 101 artificial statements, and they understood 70% of them. The results, though not perfect, but very encouraging, Dr. Chang said.

The words are made by mapping the activity of the brain that controls speech-generating movements, and then turning these movements into sounds is easier than creating sounds directly from the signal. The brain, Stephanie Riès, neurologist at the University of California, San Diego commented.

The speech synthesis device used by Stephen Hawkng requires users to spell words with eye or facial movements.(Source: The Guardian).

This latest achievement opens up the opportunity to regain the ability to communicate for people who lose their voice because of illness or injury.