How 8 seconds of a bad and noisy recording returned my voice that I lost to the disease of the motor neuron

The sound of his voice as a young man.

The “miracle” described by Sarah began when Smartbox, a Bristol headquarters -based technology assistance, asked him for an hour of audio to recreate his voice.

Sarah and her children, already adults, looked for something suitable, but Sarah had lost her voice in 2000, before smartphones became generalized and that social networks captured memorable moments.

The family found an old VHS tape recorded with a 1990 video camera in which Aviva appears when he was a baby. But the image was blurred and the distorted sound. People in the shot muttered and the audio was drowned by a television at full volume.

With difficulty, there were eight seconds of Sarah's voice.

Sarah was a Londinense Vivaz who had worked as a personal assistant in an editorial until her life took an unexpected turn.

Married and with a young son, the family waited for a second baby. Sarah began to feel bad. His speech had become slower and had weakness in his left arm.

Unexpectedly, ENM was diagnosed, sometimes known as ELA (amyotrophic lateral sclerosis).

This degenerative disease causes muscle weakness and, of the 1,000 people diagnosed in the United Kingdom every year, the NHS (Public Health Service in the United Kingdom) states that most lose the ability to speak.

The ENM affects men more and shortens life, but varies from person to person and some live a lot of time.

“I was in denial, thinking that it would be fine,” says Sarah, but “after Eric's birth, my condition worsened quickly.”

In a matter of months, he lost the use of the hands and then “all intelligible speech.” His marriage ended shortly after.

“I was very depressed and I terrified disability and death,” he says. With two young children under his charge, Sarah depended on 24 hours.

“It was difficult to see strangers take care of my children, but I am grateful,” he says, while sitting with Aviva and Eric.

Unable to move or communicate easily, Sarah fought against isolation.

He says he spent the first five years watching poor quality television, watching how his children grew.

Eric, 25, says that his only memories are “her paralyzed mother”, while Aviva, 28, recalls the moment when he realized that his mother was different.

“I just remember having asked him to prepare some strawberries, and I couldn't cut them. I had to ask someone else.”

Five years after his diagnosis, Sarah was finally able to communicate with technology of the look.

This allowed him to build words and prayers with the movement of the eyes, and speak, although with a synthetic voice, such as the physicist Stephen Hawking.

This technology allowed him to become a volunteer of the ENM association and patron of the Lifelites beneficial organization, whose objective is to provide technology to children and families so they can communicate

And he returned to his passion, painting, using the technology of the look to create original works of art.

“I was so happy, although it was arduous and exhausting,” he says.

In Bristol, Simon Poole, from Smartbox, says he shrunk his heart upon receiving only eight seconds from a VHS tape. They had asked for an hour of audio.

“I thought it was impossible to create a voice with such a bad audio,” he says.

Even so, he experienced with him, he integrated it to the most recent system of the International Artificial Intelligence Company for Voz Elevenlabs.

The company had announced that it wanted to offer free voice cloning to one million people at risk of losing speech due to conditions such as ENM, cancer or cerebral spills.

His project had worked with rugby player Rob Burrow.

Simon managed to distinguish Sarah's voice from the sound of the entire TV at full volume using the Elevenlabs voice insulator. But the result was weak, lacking intonation and personality, and with an American accent.

So he turned to another application where, using thousands of voices, AI had been trained to fill in the insulator holes and predict where a voice like Sarah and his intonation would go.

In the end, he obtained several audio phrases with which he was satisfied and sent them to Sarah.

He remembers how she told her that she almost cried when she heard her new voice for the first time. And that an old friend of Sarah, someone who knew her voice, was “impressed by how realistic it was.”

But how would Aviva and Eric react when listening to it?

“It was amazing,” says Aviva. “I am still assimilating. When listening now in my daily life, it still surprises me.”

Sarah's new voice has also joined the family, because now she can express emotions.

He has “marked a big difference,” says Eric.

“We can feel who is as a person,” adds Aviva. “Mom is not just a person with disabilities in a corner with a robot that does not identify with her.”

The voices created by AI suppose a great improvement with respect to the old computerized voices, or which could be selected from recorded voices libraries, says Susan Oman, a data specialist, AI and Society of the University of Sheffield.

“It's about you as an individual and your connection with yourself,” he says. “If that (voice) does not identify you, then you do not identify.”

The preservation of accents is also “very important” at a time when technology could homogenize them, he maintains.

“They betray your class. They betray your origin. In the world we try to recover accents and dialects that have been lost.”

Sarah jokes saying that sometimes her ancient synthetic voice misses.

“It sounded very elegant and people didn't know that it was actually a Cockney – a working class London – with a slight ceceum,” he jokes.

But she is happy to have recovered her old voice.

“I'm glad to have returned. It is better than being a robot.”

Subscribe here To our new newsletter to receive every Friday a selection of our best content of the week.

And remember that you can receive notifications in our app. Download the latest version and act.