A person attempts to speak, and a computer produces a voice. Brain–computer interface experiments use patterns of neural activity to support communication or device control when the usual movement pathway is impaired. A decoder learns relationships between measured signals and the intended action.
Signals can be recorded with scalp sensors or implanted electrodes. Each approach has tradeoffs in detail, convenience and risk. Software translates the recordings into outputs such as cursor movement or synthesized speech, often after calibration for the participant and task.
Implanted interfaces for speech restoration and complex control remain under research and clinical evaluation. A study reported by NIH in 2025 translated attempted speech into streaming audio. This is a concrete rehabilitation advance within defined experimental conditions, rather than unrestricted access to every thought. AI helps decode the signals. The connection to privacy is equally direct: the information being processed comes from the nervous system.