Brain implants convert attempted speech into a personalized voice
A new generation of brain-computer interfaces can decode attempted speech directly from neural activity, turn it into words, and present the result as synthetic speech. Unlike systems based on hand or eye movements, the approach allows users to communicate by trying to speak.
The systems can also generate a voice resembling the user’s own, giving the spoken output a more personalized character. This combination of neural decoding and voice synthesis could make speech-restoration tools more natural and less dependent on robotic or generic voices.
The approach represents progress in connecting language-related brain activity with real-time speech output, although the brief available information does not specify how the systems perform outside their tested settings or how many participants have used them.