# Neuralink VOICE Trial **Entity class:** Clinical trial **Domain:** Speech Neuroprostheses / Intracortical BCI **Doc Type:** Trial Node **Maturity:** Investigational **Collection:** [[collections/Neurotech|Neurotech]] **Primary Sources:** [[research/Research Inferences Integration Report - 2026-10-23|Research Inferences Integration Report — 2026-10-23]] ## Definition **VOICE is [[wiki/Neuralink|Neuralink]]'s investigational brain-to-voice trial, which decodes intended speech from its implant and plays it back as synthesized spoken words.** It recruits adults aged 22 to 75 with severe speech impairment and impaired upper-limb function, typically from ALS. Neuralink received FDA Breakthrough Device designation for speech restoration in 2025. The device is investigational and is not FDA approved. ## Participants and milestones - **Kenneth Shock** received his implant in January 2026 and became the second VOICE participant to demonstrate thought-to-speech output on video. - **"Terry,"** reported as the third VOICE participant, was shown in a Neuralink post on 18 September 2026 progressing from mimed speech to thinking words and hearing them spoken aloud. Neuralink stated that a Grok Voice model powers the spoken output, intended to sound like the user's own voice. - **Unpublished:** implant dates, vocabulary size, accuracy, latency (independent analysis notes several seconds), and the number of active participants. ## Composition VOICE is the first visible joining of an implant, a speech decoder and a foundation-model voice inside one founder's portfolio: Neuralink's hardware, Neuralink's decoder, and a voice model from the [[wiki/Grok|Grok]] line of [[wiki/Elon Musk|Elon Musk]]'s AI company ([[wiki/xAI|xAI]]). The restored person speaks through a trained artifact, which makes the voice itself part of the identity object the register describes. ## Forward Capability Speech restoration is converging on the same pattern across three programs: [[wiki/BrainGate|BrainGate]] home use, [[wiki/Paradromics Connexus|Paradromics Connexus]] real-time conversation, and VOICE. Neuralink's distinguishing variable is vertical integration, with the voice model supplied from inside the same capital graph. Peer-reviewed metrics will place VOICE against the published academic benchmarks. ## Open Questions - Peer-reviewed accuracy, vocabulary and latency figures. - The provenance and personalization method of the synthesized voice. - Enrollment size and sites, from ClinicalTrials.gov. ## Research Inferences These entries translate the forward-looking register in [[research/Research Inferences|Research Inferences]] into ordinary wiki prose. The tier labels apply to the inference, not automatically to every factual anchor inside it. Proposed IDs above INF-0500 and tier recommendations come from the 2026-10-23 cycle: [[research/Research Inferences Integration Report - 2026-10-23|Research Inferences Integration Report — 2026-10-23]] · [[research/Research Inferences Integration Audit - 2026-10-23|Research Inferences Integration Audit — 2026-10-23]]. - **INF-0096 — Strongly indicated** (recommended for promotion to established in the 2026-10-23 report). Restoring speech with a synthesized version of the person's own pre-morbid voice is already technically routine. The restored person therefore speaks in a voice generated from a model, and the resulting identity object is part biological, part trained artifact. - **INF-0078 — Plausible.** A decoder that emits semantic features rather than keystrokes pairs naturally with a language model that completes intent. The interface becomes a conversation between a partial neural signal and a predictive model, and most of the realized bandwidth comes from the model's prior rather than from the electrode. - **INF-0080 — Analytic.** When a model supplies most of the structure of a decoded utterance, the question of whose expression it is becomes technical rather than philosophical. Attribution ratios inside assisted communication will need to be measurable, and that measurement is itself a new instrument. - **INF-0022 — Strongly indicated.** Locked-in patients are the population for whom the interface is the only channel, which makes them the group that will use the highest-bandwidth configurations first and longest. Clinical necessity, not enthusiasm, will produce the first humans who have lived years mediated by a decoder. - **INF-0512 (proposed) — Strongly indicated.** A Neuralink speech participant speaking through a voice model from a sister company is the first visible composition of implant, decoder and foundation-model voice inside one founder's portfolio. ## Relationships - **sponsor:** [[wiki/Neuralink|Neuralink]]. - **voice_supplier (company-stated):** [[wiki/Grok|Grok]] · [[wiki/xAI|xAI]]. - **founder:** [[wiki/Elon Musk|Elon Musk]]. - **method:** [[wiki/Inner Speech Decoding|Inner Speech Decoding]] · [[wiki/Speech Restoration|Speech Restoration]]. - **population:** [[wiki/Amyotrophic Lateral Sclerosis|Amyotrophic Lateral Sclerosis]]. - **peer_programs:** [[wiki/Paradromics Connexus|Paradromics Connexus]] · [[wiki/BrainGate2 home BCI|BrainGate2 home BCI]]. - **cluster_home:** [[wiki/Cognitive Interface Layer|Cognitive Interface Layer]] · [[wiki/Brain-Computer Interfaces|Brain-Computer Interfaces]] · [[collections/Neurotech|Neurotech]]. ## Sources - [Neuralink patient Terry speaks with thoughts using Grok Voice — Teslanorth, 19 September 2026](https://teslanorth.com/2026/09/19/neuralink-terry-voice-trial-grok-voice/) (reports the Neuralink post of 18 September 2026) - [Neuralink brain implant helps man with ALS speak using his thoughts — MobiHealthNews, 2026](https://www.mobihealthnews.com/news/neuralink-brain-implant-helps-man-als-speak-using-his-thoughts) - [Neuralink device gets FDA recognition for speech restoration — Teslarati, 2025](https://www.teslarati.com/neuralink-speech-restoration-fda-breakthrough/) As of 23 October 2026.