AIBrain2voice 2.0 reportedly synthesizes voice from brain signals in real time
A researcher announcing the preprint reports a 5.24% word error rate when human listeners transcribed its output, calling it an eightfold improvement over the team's 2025 Nature results.
TLDR
A researcher describes Brain2voice 2.0 as a brain-computer interface that synthesizes voice directly from signals inside the brain's cortex in real time. Announcing the work as a preprint, they report that human listeners transcribed the voice with a 5.24% word error rate—an eightfold improvement over the team's 2025 Nature results, according to the announcement.
Brain2voice 2.0 reportedly synthesizes voice from brain signals in real time
A researcher announcing the preprint reports a 5.24% word error rate when human listeners transcribed its output, calling it an eightfold improvement over the team's 2025 Nature results.
TLDR
A researcher describes Brain2voice 2.0 as a brain-computer interface that synthesizes voice directly from signals inside the brain's cortex in real time. Announcing the work as a preprint, they report that human listeners transcribed the voice with a 5.24% word error rate—an eightfold improvement over the team's 2025 Nature results, according to the announcement.
