Neuralink has announced its latest clinical progress: Terry, an ALS patient participating in the "VOICE" speech recovery trial, successfully used the Neuralink device implanted in his body to help the research team optimize the "brain-to-voice" interface system, ultimately saying "I love you" to his wife in his natural voice, which he had lost for many years.

According to Neuralink's official introduction, Terry's training of the algorithm was divided into two phases: in the first phase, he attempted to mimic speech actions, making mouth shapes and muscle movements associated with speaking without actually producing sound; in the second phase, he only needed to "think" of the words he wanted to say, and the system could directly convert these neural signals into speech output, using a synthesized voice that resembled his natural voice before his illness. According to ABC News, Terry is the third patient to join this brain-machine interface project, and during this technical demonstration, he successfully said "I love you" to his wife for the first time using the system, which was described as an important emotional breakthrough.

The VOICE trial, officially titled "Feasibility Study of Precision Robot-Implanted Brain-Machine Interfaces for Communication Recovery," is registered on ClinicalTrials.gov (NCT07224256) and targets patients with severe speech impairments due to ALS, primary lateral sclerosis, stroke, or spinal cord injuries. It has received the FDA's Breakthrough Device Designation, which is intended to expedite the review process but does not equate to market approval. Technically, the N1 implant device collects neural discharge signals from the brain's speech planning area through hair-thin electrode wires, which are wirelessly transmitted to external devices. Machine learning algorithms then map the neural signal patterns to phonemes, which are combined into complete phrases, ultimately outputting in text or synthesized speech form.

This speech synthesis function is supported by Grok Voice, a subsidiary of xAI. Neuralink also reminded in a supporting statement that "the Neuralink device is still in the experimental stage and has not yet received FDA or other regulatory approvals," noting that the content shown in the video comes from personal experiences shared by voluntarily participating clinical trial subjects and "may not represent the situation of all participants or future treatment effects."

This is not the first time Neuralink has introduced generative AI into patient communication scenarios. Previously, participant Bradford Smith (who completed implantation in November 2024 and is the first aphasic ALS subject) had used Grok to assist in drafting replies on the X platform, and with ElevenLabs' voice cloning technology, he restored his voice before his illness; another participant, Kenneth Shock, who had his device implanted in January, demonstrated brain-controlled speech capabilities in March. Some neuroscientists have previously expressed concerns about this type of "brain-machine interface combined with generative AI" model: Eran Klein, a neurologist at the University of Washington, pointed out that such systems involve trade-offs between "expression speed" and "content authenticity," and Musk's simultaneous leadership of Neuralink, xAI, and X (formerly Twitter) has raised discussions about his ability to intervene or influence the content patients express externally.

From an industrial logic perspective, Neuralink's demonstration further solidifies the technical route of deep integration between brain-machine interfaces and generative AI. This narrative helps strengthen the valuation imagination space for Musk's two unlisted companies, Neuralink and xAI, in primary market financing—once medical-grade brain-machine interface technology is proven to be scalable and rapidly deployable, it will accumulate key clinical and public opinion capital for subsequent financing and FDA approval processes. Meanwhile, Grok Voice, as a real medical application case for voice synthesis technology, adds differentiated application endorsement for xAI in competition with similar voice and dialogue model vendors like OpenAI and ElevenLabs. In the short term, such high-profile clinical demonstrations reflect more on the accumulation of technical trust and brand volume, and have not yet directly corresponded to quantifiable commercial revenue or cash flow; their long-term capital value highly depends on subsequent FDA approval progress and validation results from larger-scale clinical data.

Source: Public Information