Could AI give this little girl a voice?


WATCH: Meet Everly, the little girl searching for her voice

Everly Paddock loves making friends. She’s five years old and attends kindergarten in Ontario, Canada. 

It’s an exciting time in her life. There are new faces, new ideas and new adventures around every corner. But making friends isn’t easy without a voice. 

Everly’s vocal cords have been partially paralyzed for most of her young life. She struggles to raise her voice above a whisper, even when she tries to shout. 

“You see it at school,” says Everly’s father, Jeff. “She’ll go up to anyone and start a conversation. Most kids will be open to that. But after a few seconds, they realize they have no idea what she’s saying. So her circle of friends is slowly moving away.”


A parent’s pain

Everly was born nearly 16 weeks premature. She could not breathe on her own, so doctors inserted a special tube down her throat. 

They hoped she wouldn’t need it for more than a few weeks. The reality was very different. 

“They would take out the tube, but then her oxygen levels would fall and you could tell she was struggling to breathe,” says Jeff. “They tried to remove it on eight different occasions, but they had to re-insert it again and again.”

Everly battled to breathe on her own after she was born 16 weeks premature. /Jeff Paddock

Everly battled to breathe on her own after she was born 16 weeks premature. /Jeff Paddock

Doctors had to perform emergency surgery after failing to reposition the tube during one of these procedures. They took the decision to insert a device in Everly’s trachea – a move which freed up her airways by bypassing the upper part of her throat. 

Everly’s voice was severely restricted while the trachea device was in place. Her parents expected that to change when doctors removed it last August. 

Instead, they got some devastating news – Everly’s vocal cords had suffered serious damage. It’s not clear when exactly this happened, but Jeff thinks it’s the result of repeated attempts to insert the breathing tube when she was a baby.   

“We were heart-broken when we were told about the vocal cord damage,” says Jeff. “I remember how long we had to wait to hear her say, ‘I love you.’ We just miss the things parents always expect to do with their children.”


A voice for the voiceless

Everly has already had seven operations to help correct the damage to her throat. Jeff is worried that another major surgery could do more harm than good. The options are limited. But there is hope on the horizon. 

Chinese scientist Dr Jun Chen has built a device which could offer a voice to the voiceless. Chen worked with his team at the University of California Los Angeles, building on technology which he developed in 2020. He thinks his ‘voice patch’ could change millions of lives. 

“This device produces speech using a system based on muscle movement,” says Chen. “I really hope this work will help people with voice disorders and vocal cord injuries.”

The ‘voice patch’ could help people to speak despite vocal cord injuries. /Jun Chen Lab/UCLA

The ‘voice patch’ could help people to speak despite vocal cord injuries. /Jun Chen Lab/UCLA

Flexing some ‘machine muscle’

People use their vocal cords to produce soundwaves and, ultimately, speech. 

But vocal cords aren’t the only part of the speaking process. People also mobilise certain muscles in their throats. Chen’s patch relies on prompts from those muscles, allowing users to speak even if their vocal cords aren’t working normally.    

Each movement generates a unique electric signal. Artificial Intelligence (AI) interprets that signal, matching it with a word in its database. The device then transmits sound in a voice that the user chooses. 

“We will use machine learning,” says Chen. “First we will build up a personalized database of words for the user. They’ll be able to wear the patch once that’s in place.” 

The patch measures just over one square inch and sticks to the skin. /Jun Chen Labs/UCLA

The patch measures just over one square inch and sticks to the skin. /Jun Chen Labs/UCLA

Chen included eight participants in the trial phase of his research. They were told to mouth several basic sentences in silence – phrases such as ‘How are you doing today?’ and ‘I love you!’ 

The results suggest the AI can identify the correct word 95-percent of the time. There is a slight delay between muscle movement and speech, but Chen says that could disappear as the technology evolves. 

“I would estimate that right now there’s a delay of about a second. Eventually users will not feel any delay. That is the ultimate goal.”


‘It would be a game-changer’

Chen says the device could be on the market in the next three to five years. That means Everly may have to wait a while before she can try it out. But her father is already gearing up to explore the possibilities. 

“I think it will be a game-changer if it works for Everly,” says Jeff Paddock. “I can’t even begin to describe how it would change daily interactions and activities. She’s at a young age, so it would make a huge impact.”

Jeff turns to Everly, a smile stretching across his face. “What would you think if we could get a ‘band aid’ that would give you a little bit more of a voice? Would you like that?” 

His daughter thinks for a second, and then responds with “Yeah.” Jeff follows up, asking “Could you give me your loudest voice?”

Everly doesn’t hesitate this time. “Yeah!” Her voice strains to get the words out; they seem to prick at her throat as they come off her tongue. But her eyes tell a story of their own. This little girl has a lot of love in her heart. 

Now, she needs a voice to share it.   

Subscribe to Storyboard: A weekly newsletter bringing you the best of CGTN every Friday