Neural Decoding of Spontaneous Overt and Intended Speech.
Debadatta Dash, Paul Ferrari, Jun Wang
PMID 39106199WHAT IT FOUND
Noninvasive brain scans accurately decoded when healthy adults spoke yes or no, and also detected the intention to speak before the words were heard.
This confirms brain-computer interface potential for speech, but the small sample and simple vocabulary mean it is not yet clinical evidence.
Key findings
01Decoding spontaneous overt speech from noninvasive brain activity was highly accurate using a convolutional neural network, significantly surpassing chance levels.
02The intention to speak was also decoded above chance levels, though accuracy was lower than for overt speech.
03The study used only two words (yes and no) in healthy adults, limiting its direct applicability to clinical populations with complex speech needs.
STILL TO COME
How it was doneWhat they foundWhat it means for SLPs
Read the rest of this summary
You get three full summaries a month, free, and we do not ask for a card. Search, the TL;DRs and your library stay unlimited either way.
What it does not show
The study included only seven healthy participants. The vocabulary was restricted to only two words ('yes' and 'no'). Decoding of intended speech may have been influenced by motor artifacts (muscle movement) preceding the actual speech, rather than pure neural intention. The findings have not been validated in clinical populations such as those with ALS or locked-in syndrome.
Declared interests
Funded by the National Institutes of Health (NIH) and the University of Texas System. The authors declared no other conflicts of interest.
The easy way to misread this
Do not assume this technology is ready for patients with communication disorders. The study used healthy adults and only two words, and the 'intended speech' decoding may have partially detected muscle movement rather than pure cognitive intention.