South Korean POSTECH builds neckband converting silent mouthing into speech

Technology Apr 24, 2026

A team at South Korea's POSTECH has built a smart neckband that can turn silent mouthing into actual speech.
The lightweight device uses a silicone base, tiny camera, and motion sensors to pick up on subtle neck muscle movements.
With an AI model trained on your own voice, it can generate words even if you don't make a sound.

POSTECH neckband achieves 85.8% NATO accuracy

Tested with the NATO phonetic alphabet, the neckband hit 85.8% accuracy across 26 words and worked reliably even in loud environments.
It only takes about 10 minutes to train it on your voice.
The team hopes this tech could help people with speech disorders or anyone who needs to communicate quietly, like in noisy places or when using a mic isn't an option.