A device has been created by scientists that can translate speech ideas into spoken words in real time.

Although still in the experimental stage, the goal is to develop a Brain Computer Interface that can give voice to individuals unable to speak.

In a recent study, the device was tested on a 47-year-old woman with quadriplegia who had been speech-impaired for 18 years since experiencing a stroke. The device was implanted in her brain during surgery as part of a clinical trial.

According to Gopala Anumanchipalli, co-author of the study published in Nature Neuroscience, the device “translates the intent to speak into fluent text.”

Most brain computer interfaces for speech experience a delay between thought and speech, which can disrupt conversations and cause misunderstandings. However, this new device is considered a significant advancement in the field.

The device works by recording brain activity using electrodes and generating speech based on this activity. An AI model is then trained to translate this neural activity into spoken words.

Anumanchipalli of the University of California, Berkeley, explains that the device operates similarly to existing systems used for transcribing meetings and phone calls in real time.

Located in the brain’s speech center, the implant translates signals into spoken sentences as they are heard. This “streaming approach” ensures a constant flow of audio to the recorder without waiting for the sentence to finish.

Rapid speech decoding enables the device to keep up with natural speech pace, enhancing language naturalness according to Brumberg.

Funded in part by the National Institutes of Health, further research is necessary before the technology can be widely available. Anumanchipalli suggests that with sustained investment, the device could potentially be accessible to patients within the next decade.