- AI used to convert neck movements into words
- It is the work of researchers from South Korea.
- The system could eventually be used for ‘silent conversations’
One of the many talents that AI systems offer is advanced pattern recognition, and researchers at Pohang University of Science and Technology (POSTECH) in South Korea have taken advantage of this by developing a wearable neckband that translates muscle and skin movements into speech.
As described by the researchers (via New Atlas), the device is intended to restore “lost voices”—that is, return the power of speech to those who cannot speak normally. A small camera is used to track a “motion map” on the neck when the user tries to say something, without the need for vocalizations.
Algorithm-based estimation of what a person is trying to say is then combined with speech synthesis technology to verbalize the words. This speech synthesis system can even be trained on the AI neckband wearer (an example of deepfake technology actually being used for something worthwhile).
Article continues below.
“It is a noteworthy technology because it has a wide range of potential applications, including assisting laryngectomized patients, communicating in noisy industrial environments, and even supporting quiet conversations,” says Professor Sung-Min Park, who led the study at POSTECH.
85.8% recognition accuracy
Such “silent conversations” could take place in libraries or conference rooms, for example, when people cannot speak audibly for whatever reason. It could even be used as an alternative to radios or text messages for staff working in an area where face-to-face conversations are not practical.
However, there is a long way to go before this is ready for anyone to use. While the band can achieve a recognition accuracy of 85.8%, this occurs only on 26 predefined words: the NATO phonetic alphabet (“Alfa”, “Bravo”, etc.). Meanwhile, when the subject is moving, the accuracy rate currently drops to 39.72%.
That rate should improve as the system becomes more capable. Encouragingly, the neckband worked well at eliminating background noise, and words could be recognized with white noise levels of 90 decibels in the background (that’s roughly the equivalent of a busy construction site).
Another benefit of the wearable AI system is that it is much simpler and less bulky than other think-to-speech systems currently in development, including technology based on electroencephalography (EEG) brain wave readings. It is an important step in terms of portability and accessibility.
Follow TechRadar on Google News and add us as a preferred source to receive news, reviews and opinions from our experts in your feeds. Be sure to click the Follow button!
And of course, you can also follow TechRadar on YouTube and tiktok for news, reviews, unboxings in video form and receive regular updates from us on WhatsApp also.
The best laptops for all budgets
