Researchers at the University of California, Davis, have developed an experimental brain-computer interface (BCI) that restores real-time conversational ability to individuals who have lost the ability to speak due to neurological conditions.
This technology translates brain activity into voice, effectively creating a digital vocal tract without noticeable delay. The BCI utilizes artificial intelligence to decode the patient's brain's electrical activity during speech attempts.
The device reproduces not only the intended words but also natural speech characteristics like intonation, tone, and emphasis. This allows for a more natural and expressive form of communication.
The system allowed a participant with amyotrophic lateral sclerosis (ALS) to "speak" through a computer, change intonation, and "sing" simple melodies. The synthesized voice was often intelligible, with listeners correctly understanding almost 60% of the words.
The process is facilitated by advanced AI algorithms trained on data collected while the participant attempted to pronounce phrases. The algorithms map neural activity to desired sounds, enabling the synthesis of speech nuances and giving the participant control over their BCI voice cadence.
The technology enabled the participant to say new words and make interjections. The study, published in Nature, represents a significant advancement over previous BCI models.