Brain-Computer Interfaces
Radbounds University and UMC Utrecht researchers achieved a remarkable milestone in the field of Brain-Computer Interfaces. It uses brain implants and AI to decode the brain signals to translate them into audible speech. The accuracy rate is also found to be ranging from 92% to 100%.
This sounds very remarkable because this might hold the potential to give paralyzed individuals a way to speak and get their voice back in other forms. A study published in the Journal of Neural Engineering also made a huge leap in BCI technology.
Views & opinions from others
Berezutskaya put forth his words, “Ultimately, we hope to make this technology available to patients in a locked-in state, who are paralyzed and unable to communicate. These people lose the ability to move their muscles, and thus to speak. By developing a brain-computer interface, we can analyze brain activity and give them a voice again.”
One of the things to note is, that this research is based on a relatively small dataset. Even with that, the model uncovered the intricate connection between brain activity and speech. Imagine the same, when it is fed an optimum dataset, the reconstructed speech might even resemble the tone of voice and manner of speaking as well. One can read the entire research here.
Limitations
Currently, there are limitations to be addressed such as volume. It focuses on predicting words individually and not sentences. It is less complex than predicting entire sentences as a whole. For sentences, more extensive experience, advanced brain implants, datasets, and refinements are needed.
The road ahead is long for sure, but further continued research might take it in the right direction. Thus, this research marks an amazing milestone in developing BCI’s ability to restore speech and voice for individuals who are no longer capable.
With more robust research and technologies, there might soon be a day when it can do sentences than just words. It will not just revolutionize the speech but carry forward into other wide applications as well.