Using recent advances in AI-based modeling, researchers developed a neuroprosthesis that synthesizes brain signals into audible speech in near-real time. The advance enables a more naturalistic speech pattern for people unable to talk because of paralysis.