Experimental brain implants at UC Davis decoded neural signals from a 46-year-old man who lost his voice to ALS and translated them directly into more intelligible speech, and helped a 58-year-old woman who is a trained singer produce a musical scale, The Mercury News reported.
The work is scheduled to be presented next week at the Society for Neuroscience conference in Washington. Cindy Chestek, a University of Michigan biomedical engineer who was not involved, told the paper, "This system, for the first time, is good enough. Good enough for real communication. It's good enough to sing."
Earlier brain-computer interfaces often turned attempted speech into text first, then into a computer-generated voice after a patient mentally hit a kind of "send" button. That delay made true conversation difficult. "Imagine you can only speak just by texting for the rest of your life — that's not sufficient, right?" said Maitreyee Wairagkar, a neuroengineer at UC Davis's Neuroprosthetics Laboratory who led the work on the two patients.
One participant, Casey Harrell, a climate activist, began the study in 2023 at age 45, when four arrays totaling 256 electrodes were implanted in his brain. He has used the system to generate nearly 2 million words over two years, including conversations with family, messages to friends, emails and work on a laptop, the paper reported. In 2025 the team published in Nature a demonstration of direct brain-to-voice speech with about a 10-millisecond delay, but listeners could accurately transcribe only a bit more than half of his spoken words. In the new work, described in a preprint, researchers fine-tuned the algorithms on the same dataset and increased word intelligibility by up to 94%.
"I think this is really clinically meaningful improvement and not just impressive machine learning," said Christian Herff of Maastricht University, who was not involved. Mariska Vansteensel of University Medical Center Utrecht praised the work but said that because it reanalyzed previously collected data, it will be important to see how well the improved decoding works in real time and in people with different degrees of speech loss. She also noted that systems useful outside a lab will need to be fully implantable; current setups still tether participants to computers through ports in the skull. Companies such as Neuralink and Paradromics are testing wireless systems, she said, but those results are still being shown.
In a second experiment from the same team, a participant known as T21, who could no longer speak intelligibly when she enrolled, had 384 electrodes implanted. She attempted to sing and to modulate tone, controlling pitch differences that can turn a statement into a question or change the emotion of a line. As speech generation has improved, research groups have also started to focus on other parts of human communication, including gestures and tone, the Mercury News reported.
The UC Davis results do not mean a widely available device is ready for the clinic. They do show a California lab pushing brain-to-voice systems from delayed text into something closer to conversation, and even toward music, for people whose illness has taken their speech.