AI and a Faded VHS Tape Restore a Voice Lost to MND

In a powerful testament to the advancements in artificial intelligence and a mother's enduring spirit, a woman who lost her voice to Motor Neurone Disease (MND) over two decades ago has had it restored using a mere eight-second, scratchy audio clip from an old videotape. The story of Sarah Ezekiel is not just a triumph of technology; it is a profoundly human tale of a life forever changed by a devastating diagnosis, and the incredible journey to reclaim a fundamental part of her identity.
Diagnosed with MND at the age of 34, shortly after the birth of her second child, Sarah Ezekiel experienced the rapid and relentless progression of the neurological condition. Within months, she had lost the ability to use her hands and, most devastatingly, all intelligible speech. Her two children, Aviva and Eric, have no memory of hearing their mother's natural voice. For their entire lives, they have only known her through a machine-generated voice, a monotone, robotic tone that, while a lifeline for communication, lacked emotion, inflection, and the unique identity of her real voice. This loss of her voice created a profound sense of isolation, making it difficult for her to express emotion—to show happiness, sadness, or anger—and to feel fully connected to her children and the world around her.
For years, Sarah communicated using eye-gaze technology, which allows her to type by tracking her eye movements across a screen. While this technology has been transformative for many people with MND, it still produced a synthetic voice, a substitute for the real thing. Then, an opportunity arose. A Bristol-based assistive technology company, Smartbox, which specializes in creating personalized communication systems, offered to recreate her voice using AI. The process typically requires at least an hour of good-quality audio to capture the nuances of a person's speech and create a realistic AI voice. For Sarah, this presented a monumental challenge. Having lost her voice before the widespread use of smartphones and digital recordings, she had almost no audio of herself speaking.
The breakthrough came from an unlikely source: a forgotten family videotape from the 1990s. The tape, a relic of a different era, contained a muddled, barely audible, eight-second clip of Sarah chatting in the background. The audio was filled with background noise, including a blaring television and other people talking. It was a fragment of a conversation, a tiny and seemingly insignificant moment from a life long past. Despite the slim chances of success, Smartbox's audio experts, led by a sound designer, took on the challenge. They meticulously cleaned and enhanced the scratchy audio using a variety of specialized tools, working to isolate and rebuild her voice from the minute fragments available.
The result of their painstaking work was a stunning success. For the first time in over two decades, Sarah heard her own voice again—complete with the accent and a slight lisp she used to hate. It was a deeply emotional moment. "After such a long time, I couldn't really remember my voice. When I first heard it again, I felt like crying. It's a kind of miracle," she said, communicating through her eye-gaze technology and her new AI voice. Her children's reaction was equally moving. For them, hearing their mother's voice was a revelation. It allowed them to connect with her on a new level, to hear the personality and emotion they had only ever imagined. Her son, Eric, noted that her new AI voice has brought the family closer, as his mother can now truly express her feelings. Her daughter, Aviva, described the experience of hearing her mother's voice as "amazing," stating that it helped her to "feel who she is as a person."
This incredible story is not just a one-off. It is part of a broader movement to use AI and assistive technology to help people with degenerative conditions preserve and reclaim their voices. Organizations are now actively "voice banking," encouraging people with MND to record their voices early in their diagnosis so that a high-quality digital replica can be created before they lose the ability to speak. The case of Sarah Ezekiel, however, highlights the transformative power of the technology even for those who did not have the opportunity to plan ahead. It demonstrates that with the right tools and a commitment to innovation, even the smallest traces of a person's past can be used to restore a fundamental part of their identity. Her story offers a powerful message of hope and reminds us of the profound connection between voice and self.
Source@BBC