In a remarkable leap forward for communication technology, scientists have developed a brain-to-voice device that allows people with severe paralysis to speak using only their thoughts. The neuroprosthesis, created by researchers at UC Berkeley and UCSF, decodes brain activity into audible speech nearly instantly thereby eliminating the long delays of earlier systems. Using artificial intelligence, the model interprets signals from the motor cortex—where speech movements originate—and streams words aloud with natural rhythm and timing.
The system was tested with a participant named Ann, who lost the ability to speak after a brainstem stroke. By silently attempting to say phrases, Ann’s brain activity was transformed into audio that sounds like her pre-injury voice. Unlike earlier systems that took seconds to process each sentence, the new model responds within one second, enabling more fluent communication. Researchers say this advance brings them closer to fully restoring expressive, real-time conversation for people who cannot speak.
And for Ann, hearing her voice again wasn’t just functional—it felt like reclaiming a part of herself.