Researchers from the University of California have developed an implant which can decode the brain signals to give speech output.
The scientists did tests on five volunteers who read passages from children’s novels. They used an electrode to measure the voltage across the brain and then those readings were matched to produce the speech output.
This attempt is different from the previous ones. Instead of interpreting the brain’s thoughts, the scientists read the brain’s efforts to produce movement in the lips and larynx.
This new device had a success rate of over 69% which is very impressive for this first-generation device. This relationship between the brain and vocal tract is especially complex and thus scientists thought of this approach.
Using Deep learning, they created two neural networks. One of them worked to interpret the movements in the throat and the other one to decode this into speech.
The new device has shown incredible progress especially when it comes to interpreting complex messages and sentences. This entire research will be sped up if high-density electrodes, powerful computing systems, and stronger algorithms are used.
Using different participants and then cross-referencing each of their responses, they get various overlaps which then helps to enhance the entire system.
The above-noted effort might not be even remotely close to the actual speech density and pace. But it is still a huge leap forward in the right direction. If successful, this device will truly change the lives of the people who are impaired.