A new brain-machine interface can translate the gist of what a person hears

Neuroscientists from the University of Texas at Austin developed a "language decoder" that captures the gist of what a person hears based on their brain activation patterns.

The method doesn’t require surgery, but uses functional magnetic resonance imaging (fMRI) to generate brain maps that correspond to langu the University of Texas also Austin late imagined speech, and even generate descriptive subtitles for silent movies using neural activity.

Neuro Translator can translate v one al information directly into a narrative based on a representation of language encoded in brain activity.

The technology only works after months of careful brain scans in a machine while lying still.

The decoder only works with the volunteers’ active mental participation.

The team is working on translating the technology to fNIRS (functional near-infrared spectroscopy). Huth Huth months Jerry Tang first seven Alexander Huth > one-minute-long one this week aryF Nature Neuroscience NxlGi">fidelity first class="summaryFeed_highLightText__NxlGi">fed Huth one A Neuro Translator Comparing AI fed one Bayesian secondHuth Decoding Language Translating ghtText__NxlGi">Reddit One ss="summaryFeed_highLightText__NxlGi">over 200 million GPT GPT Huth ightText__NxlGi">first fed noisy magnet seven Huth htText__NxlGi">Two One two BOLD ryFeed_highLightText__NxlGi">one BOLD ghLightText__NxlGi">Huth one