Meta has published study about the “brain decoder”, which, with the help of artificial intelligence, converts thoughts into speech.
According to the developers, their method is based on non-invasive methods of recording brain activity. The researchers applied electroencephalography (EEG) and magnetoencephalography (MEG) using external sensors.
With their help, the team collected over 150 hours of recordings from 169 healthy volunteers to train the algorithm. The developers noted that EEG and MEG are less reliable than sensors implanted in the brain. Therefore, they needed to collect more data to improve the accuracy of the AI model.
The algorithm recorded the reaction of the participants’ brains to audio books and individual phrases in English and Dutch. Then he isolated the necessary words from the text and compiled a “dictionary”, according to which they launched the reverse process – deciphering thoughts and converting them into text.
According to the researchers, the accuracy of the algorithm reached 73% when using a set of 793 words frequently used in everyday life.
Scientists believe that their development will help not only millions of people who have lost the ability to speak and write, but also in the study of the human brain.
In the future, the researchers plan to increase the algorithm’s initial vocabulary so that it can more accurately identify words.
Recall that in August, Meta introduced a chat bot with 175 billion parameters. Less than a week after the release of the virtual interlocutor was convicted of anti-Semitism and dissatisfaction with Facebook.
In July, researchers at Meta developed the Sphere AI algorithm for fact-checking Wikipedia.
In the same month, the tech giant introduced the NLLB-200 artificial intelligence model for online transfers. The algorithm supports 200 languages, including less common ones.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
Meta has published study about the “brain decoder”, which, with the help of artificial intelligence, converts thoughts into speech.
According to the developers, their method is based on non-invasive methods of recording brain activity. The researchers applied electroencephalography (EEG) and magnetoencephalography (MEG) using external sensors.
With their help, the team collected over 150 hours of recordings from 169 healthy volunteers to train the algorithm. The developers noted that EEG and MEG are less reliable than sensors implanted in the brain. Therefore, they needed to collect more data to improve the accuracy of the AI model.
The algorithm recorded the reaction of the participants’ brains to audio books and individual phrases in English and Dutch. Then he isolated the necessary words from the text and compiled a “dictionary”, according to which they launched the reverse process – deciphering thoughts and converting them into text.
According to the researchers, the accuracy of the algorithm reached 73% when using a set of 793 words frequently used in everyday life.
Scientists believe that their development will help not only millions of people who have lost the ability to speak and write, but also in the study of the human brain.
In the future, the researchers plan to increase the algorithm’s initial vocabulary so that it can more accurately identify words.
Recall that in August, Meta introduced a chat bot with 175 billion parameters. Less than a week after the release of the virtual interlocutor was convicted of anti-Semitism and dissatisfaction with Facebook.
In July, researchers at Meta developed the Sphere AI algorithm for fact-checking Wikipedia.
In the same month, the tech giant introduced the NLLB-200 artificial intelligence model for online transfers. The algorithm supports 200 languages, including less common ones.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
