Scientists have achieved a major breakthrough in brain-computer interface (BCI) technology by decoding inner speech — the silent voice in a person’s head — with accuracy rates as high as 74%. The study, published August 14 in Cell by researchers at Stanford University, suggests that imagined speech may soon allow people with severe paralysis to communicate quickly and naturally without speaking aloud or moving at all.
The team implanted microelectrodes in the motor cortex — the part of the brain responsible for speech — of four participants with amyotrophic lateral sclerosis (ALS) or brainstem stroke. Participants were asked to either try speaking or silently imagine specific words. Both actions activated similar brain regions, but inner speech produced weaker signals.
Artificial intelligence models trained on these neural patterns successfully decoded silently imagined words from a vocabulary of up to 125,000 words. The researchers also developed a password-based control system, enabling participants to unlock the decoder only when silently repeating a chosen phrase, such as “chitty chitty bang bang,” which the system recognized with more than 98% accuracy.
Current BCIs can translate attempted speech into text by detecting movement-related brain signals, but this approach can be slow and exhausting for people with limited muscle control. Decoding inner speech directly could allow faster, less effortful communication.
Although today’s systems cannot yet decode free-form inner speech without errors, advances in sensor resolution and machine-learning algorithms could soon make seamless, thought-to-speech interaction possible. According to senior author Frank Willett, this work lays critical groundwork for restoring natural communication to people who cannot speak, offering “real hope that future BCIs will one day feel as fluent and effortless as conversation.”
