Artificial intelligence ‘gives back the voice’ to a woman who stopped speaking 18 years ago due to a stroke

A teacher who stopped speaking at the age of 30 can communicate again 18 years later and with her voice thanks to artificial intelligence (AI).

Teacher Ann suffered a stroke while playing volleyball. This left her paralyzed at the age of 30, Univisión News reported.

Until recently, according to the news channel, he could only communicate via a letter board.

Pants made to help paralyzed people walk

Artificial intelligence helped her

The AI ​​gave her back her voice nearly 20 years after the stroke, Univisión reports, and the voice given to the woman comes from a video shot on her wedding day, in which she directs a few words.

“What time are you coming home?” the man heard her ask, as they both looked intently at a TV screen.

“It was an emotional moment to hear his voice again,” a researcher told Univision News.

“We used an excerpt from the video of her wedding to give her back the voice she had before the stroke,” explains one of the doctors from the team who managed to give Ann another chance with AI to the American chain.

Some of those behind this breakthrough are from the University of San Francisco, California.

The team implanted a very thin rectangle of 253 electrodes on the surface of Ann’s brain, in areas essential for speech.

researchers

“They discovered how to use Ann’s brain signals and translate them into words, movements and facial expressions, using artificial intelligence,” explains Univisión.

The researcher continued: “There are electrodes on the surface of Ann’s brain and when she tries to move her mouth, we decode that activity into sounds and movements that the avatar repeats.”

Advances in AI allow them to synthesize speech with the avatar, the outlet stressed.

4 unique human traits that are impossible for artificial intelligence

The most scientifically authoritative journal Nature explained what they did with Ann: “They placed a wafer-thin rectangle containing 253 electrodes on the surface of the cerebral cortex. The technique, called electrocorticography (ECoG), is considered less invasive and can record the combined activity of thousands of neurons simultaneously.

The team “trained artificial intelligence algorithms to recognize patterns in Ann’s brain activity related to her attempts to pronounce 249 sentences using a vocabulary of 1,024 words.”

“The device produced 78 words per minute with an average word error rate of 25.5%.”

Ann was able to address the researchers in a feedback session after the research. Nature commented, “When I was able to stand up for myself, it was huge!” (JO)

Questions and answers from the 2023 Elections