A new device helped survive after a stroke to talk again after 18 years
Although it is still experimental, the inventors hope that the brain-computer interface one day will help to give voice to people who cannot speak.
The new study describes the testing of the device on a 47-year-old woman with quadriplegia who has not been able to speak for 18 years after a stroke. Doctors implanted the device into its brain during surgery during a clinical trial.
She « turns her intention to speak in smooth sentences, » says Gopala Anumanchipali, co -author of the study published in Nature Neuroscience.
The power of thought: a paralyzed man walk through implants
Other brain computers usually have a slight delay between thinking about sentences and computer verbalization. Such delays violate the natural course of the conversation, which potentially leads to misunderstanding and disappointment, the researchers say.
This is « quite a lot of progress in our area, » says Jonathan Bruberg of the Laboratory of Speech and Applied Neurology at the University of Kansas, which did not participate in the study.
The California team has recorded a woman’s brain activity with the help of electrodes as she utters sentences without sound in her brain. Scientists have used a synthesizer they designed to use her voice before disability to create the speech sound she would say. They trained a model of artificial intelligence that « translates » neural activity into sound units.
It works similar to existing systems used to transcribe meetings or telephone calls in real time, says Anumanchiples of the University of California in Berkeley.
Patient with Parkinson’s passes again with neuroprosthesis after 25 years
The implant itself is placed in the speech center of the brain so as to « eavesdrop », with these signals being transformed into parts of speech that make up sentences. According to Anumanchipali, this is a type of « streaming », with every 80 -mm part of the speech – about half a syllable – is sent to the registering device.
« Don’t wait for the sentence to end – the device processes it on the go, » Anmanchipali explained.
Decoding the speech so quickly in real time has the potential to keep up with the fast pace of natural speech, Bruberg said. The use of voice samples of the patient would be significant progress in the naturalness of speech, the scientist summarized.
