Could machine learning give a literal voice to the voiceless?
Many people suffer from voice disorders. Some diseases target vocal cords. Throat cancer might require surgeries that remove part or all of the larynx (voice box). These conditions can make speech difficult or impossible. Healing—if possible—requires lengthy therapy or invasive surgery.
But what if it didn’t?
Bioengineers at UCLA invented a patch that could let people speak without using vocal cords. The thin, one-inch-square patch doesn’t require surgery or an invasive operation. It simply sticks to one’s throat like tape. From the outside, it looks like a grid of small black squares. On the inside, it holds cutting-edge technology.
This bioelectric system was invented by Jun Chen. He’s assistant professor of bioengineering at the UCLA Samueli School of Engineering.
“This new device presents a wearable, non-invasive option capable of assisting patients,” says Chen.
The device uses a technology called magnetoelasticity. Try saying that five times fast! But you can break it down into its parts: magneto and elastic. Magnetic parts push together and pull apart in a stretchy, elastic material. This movement generates power. When attached to a person whose muscles move regularly, the device can power itself!
But that doesn’t explain how this patch turns silence into speech. For that, you need machine learning.
Machine learning, or “artificial intelligence,” can detect and recreate patterns. Muscles in the throat move to create speech. Machine learning can figure out which muscle movements go with which sounds.
The neck patch contains two main components. The first component picks up muscle movements and turns them into electric signals. The second component turns those signals into speech.
To train the machine, researchers placed the new patch on eight healthy participants. The device picked up data while the participants spoke certain words. It learned the muscle patterns while they spoke. Soon, it could figure out how to produce the right sounds based on muscle movement.
The participants spoke test sentences such as, “Hi, Rachel, how are you doing today?” and “I love you!” They said the sentences both aloud and silently. With nearly 95% accuracy, the machine could match the sound!
Researchers plan to keep training the device to learn more and more sounds. Soon, they will test it on people with actual speech disorders.
God heals people in all sorts of ways. He doesn’t always do it with miracles. Sometimes, he uses the skill and creativity of people like Chen. How is God using you?
Why? New technology applied well can bring creative solutions and mercy to old problems in many areas of life’s challenges.