People can speak without vocal cords thanks to a new AI-assisted wearable device | UCLA

Content introduction

UCLA bioengineers have developed a breakthrough wearable device that converts throat muscle movements into speech, potentially restoring the ability to communicate in people with vocal cord dysfunction. The thin, flexible patch is worn around the neck and uses machine learning to achieve nearly 95% speech conversion accuracy. The technology is non-invasive and represents a significant improvement over current methods such as electrolaryngeal instrumentation or surgery, which can be inconvenient or uncomfortable. The device has the potential to help those recovering from throat cancer surgery or those suffering from speech disorders, a major leap forward in medical technology that provides affected individuals with a new level of independence and quality of life.

Automatic summary

– UCLA bioengineers have developed a wearable device that could help people with voice disorders regain their ability to speak.
– The device is a thin, flexible patch that is placed on the neck and converts throat muscle movements into audible speech.
– Through machine learning training, the device can identify which muscle movements correspond to which words, with an accuracy of nearly 95%.
– The device is non-invasive and can be used by patients with vocal cord problems or those recovering from throat cancer surgery.
– Voice disorders affect nearly 30% of the population, and current solutions may be inconvenient or uncomfortable.
– The device consists of two components: a self-powered sensing component and a driving component.
– The sensing component detects and converts muscle movement signals into electrical signals, which are then converted into speech signals using machine learning algorithms.
– Driver components convert speech signals into audible vocal expressions.
– The device is small and lightweight and can be easily applied to the throat using biocompatible tape.
– In experiments, the device demonstrated an overall prediction accuracy of 94.68% in identifying laryngeal movement signals and matching them to corresponding sentences.
– The research team plans to continue expanding the device’s vocabulary through machine learning and test it on people with speech disorders.

Original link: https://newsroom.ucla.edu/releases/speaking-without-vocal-cords-ucla-engineering-wearable-tech