A patient with ALS used a brain-computer interface to decode imagined finger movements into musical notes — composing and performing an original piece live before an audience.
In March 2026, a 54-year-old musician diagnosed with ALS — a progressive neurological disease that had left him almost completely paralysed — performed an original piano composition to a live audience at the Royal Festival Hall in London. He did not move. There was no keyboard in front of him. The music came from his imagination, decoded by a chip implanted in his motor cortex and translated in real time by an AI model trained on his own neural signals.
The participant spent six months training the neural decoder. In sessions lasting 90 minutes, he imagined playing specific keys on a piano while researchers recorded his neural firing patterns via the implant. A recurrent neural network learned the mapping between those patterns and the intended notes. Over time the model's accuracy improved from around 60% to the final 94.2% — sufficient for fluid real-time performance. The AI also learned to distinguish between imagined notes and actual compositional intention, filtering out exploratory mental "drafts."
"I have not lost music. They took my hands. They did not take my music."
— Study participant, Royal Festival Hall, March 2026