AI implants restoring speech
AI implants restoring speech, or how the brain-computer interface (BCI) works
BCI research
Challenges and hopes for the future
AI implants restoring speech, or how the brain-computer interface (BCI) works
For decades, scientists have been working on technologies that will help paralyzed people communicate with the outside world. In recent years, advances in artificial intelligence (AI) have significantly accelerated this research. This is leading to new brain-computer interfaces (BCI) with unprecedented accuracy and speed.
Groundbreaking research gives hope to people who have lost the ability to speak. New artificial intelligence-based implants translate brain signals into speech and facial expressions of a digital avatar. They do it with unprecedented accuracy and speed. Previous brain-computer interfaces (BCIs) were frustratingly slow, limiting communication to 14 words per minute. The new BCIs decode speech at 62 and 78 words per minute, approaching the natural conversation rate (approximately 160 words per minute). These new BCI implants work by translating brain signals into speech or text and the facial expressions of a digital avatar. This means that people who have lost the ability to speak due to accidents, strokes or neurodegenerative diseases can again communicate with their surroundings in a natural and intuitive way.
BCI Research
Francis Willett’s team at Stanford University has developed a BCI that interprets neural activity at the cellular level and translates it into text. Pat Bennett, a patient with amyotrophic lateral sclerosis, tested the BCI with a limited vocabulary (50 and 125,000 words). In the first case, BCI performed 3.4 times faster than earlier models, with an error of 9.1%. In the second case, the error increased to 23.8%, but still 3 out of 4 words were interpreted correctly.
Edward Chang’s team from the University of California, San Francisco used electrocorticography, recording the activity of thousands of neurons. Patient Ann, who lost her speech after a stroke, tested the BCI with a vocabulary of 1,024 words. The device generated 78 words per minute with an error of 25.5%. Additionally, algorithms transformed Ann’s brain signals into a voice and an animated avatar with facial expressions. The voice was personalized based on recordings from the patient’s wedding.
Challenges and hopes for the future
Further improvements are needed before BCIs are available for clinical use. Scientists are striving to create wireless devices with higher decoding speeds and accuracy. BCIs also need to be tested on larger numbers of people. Despite the challenges, new BCIs offer real hope for restoring smooth communication to paralyzed people. Faster and more accurate interfaces can open doors to their world, allowing them to freely express their thoughts and feelings, maintain relationships and continue working.
