New Technology Allows Man With Speech Loss to Communicate In Real-Time Using Only His Brain

On July 14th, Facebook announced in a blog post that their University of California, San Francisco research collaborators have been able to successfully develop a “speech neuroprosthesis”, which allows a man with severe speech loss to communicate in full sentences in real-time, by converting signals from the motor cortex in his brain to the vocal tract, directly into text on a screen.

The man, after suffering a series of strokes, had paralysis in his limbs and vocal tracts. Before this breakthrough, he needed to use a heavy, inefficient, head-mounted apparatus, to type what he wanted to say. However, this discovery means that now he only has to attempt to speak, and the computer can translate what he says, making communication significantly easier.

In July 2019, UCSF published the initial results, which showed that a small number of spoken words were able to be translated from neural activity. This was followed by a breakthrough full-sentence decoding in 2020 using a neural network. However, this breakthrough marks the first time an individual with severe speech loss has been able to communicate instantly, just by attempting to speak.

Dr Edward Chang, Chair of Neurosurgery at UCSF, has been “working on this goal for over a decade”, and hailed “advances in machine learning in the last 5 years” and the help of Facebook as factors which have enabled this achievement.

If anything, this achievement demonstrates the astounding potential for improving people’s lives that AI has, as well as increasing understanding of how to compute the brain - leading us one step closer to a future where we can use our brains as an interface for a computer.