Artificial Intelligence Turn Brain Activity Into Speech
How does artificial intelligence turn brain activity into speech? Many people who are paralyzed and unable to speak have signs of what they want to say hidden in their brains. Unfortunately, no one was able to decipher these signals directly. However, three research teams have recently made progress in converting data from surgically placed electrodes in the brain to computer-generated speech. Artificial intelligence-based computer models known as neural networks can rebuild vocabulary and phrases that were sometimes clear to human listeners.
In recent months, none of these efforts described in research has been able to recreate the speech that people barely imagined. Instead, researchers observed parts of the brain as people read aloud, and spoke silently. Or listened to specific recordings.
People who have lost the ability to speak after a stroke can use their eyes, make small movements to control the cursor or choose letters on the screen (as in cosmologist Stephen hawking. A small sensor mounted on his glasses is placed to take orders by tightening cheek muscles). However, using the median mind computer. They may regain control of the sound’s place and tone or the ability to get into quick conversations.
The pattern of Neurons in Artificial Intelligence
“The obstacles are many, as we try to discover the pattern of neurons raised and stopped at different time points and infer speech sounds, but the planning varies from person to person,” said Nima Mesgarani. A computer scientist at Columbia University. Likewise, the translation of these signals varies from person to person. It takes to open the skull to give models better results with very accurate data.
In rare cases, researchers do such an aggressive act, one while removing a brain tumour. As electrical readings from the exposed brain help surgeons identify and avoid the key to speech and motor areas. The other case is when electrodes are implanted in a person with epilepsy for several days to determine the origin of these seizures before surgery. “We have a maximum of 20 minutes and maybe 30 minutes to collect data, and our time is minimal,” Martin says.
The aggregates behind the new research created the most valuable information by feeding information to neural networks that address complex patterns by passing data through computational layers (nodes). Networks learn by adjusting communication between nodes.
Computer Scientists – Artificial Intelligence
Computer scientists Miguel angrick and Christian hero. Who works with the University of Maastricht, trained a network that developed specific readings of audio recordings. And then reconstructed words from brain data, previously invisible
Finally, neurosurgeon Edward Chang and his team at the University of California, San Francisco, reconstructed a complete sentence of three patients with epilepsy who had read aloud. As brain activity was captured from speech and movement centres.
166 people had to choose one of the ten written choices they heard. Some sentences were correctly identified more than 80% of the time. The researchers did more, reinventing sentences by recording data for people who read silently, harf says, “that is, to be one step away from the “alternative speech-speech” prosthesis that goes on in our mind. And this is an important outcome.”
However, Stephanie Reese, a neuroscientist at the University of San Diego in California who teaches language production, says, “What we’re waiting for is how these methods will interact with patients who can’t talk,” signals the brain when someone speaks silently. Hears their voice wandering their head. And is not identical to speech or hearing signals.
Without the sound matching speech activity. It can be difficult for the computer even to sort out when and when to start talking. Harf says one of these methods may be to give feedback to the user of the median mind computes. If they can listen to the interpretation of speech on the computer in real-time. They may be able to modify their ideas to get the results they want. With enough training for both users and neural networks program. The brain and computer may meet in the middle.