Сара Езекіль та її дорослий син Ерік/AFP
British artist Sarah Ezekiel has been deprived of the ability to speak in her own voice for 25 years due to a motor neuron disease.
However, now she can hear her own voice again, thanks to artificial intelligence and a barely audible 8-second excerpt from an old home video. At the age of 34, Sarah has Ezekiel was diagnosed with amyotrophic lateral sclerosis. At the time, she was pregnant with her second child.
This disease leads to the gradual destruction of parts of the nervous system, which can cause weakness of the muscles of the tongue, mouth and throat, which makes some patients physically unable to speak with their own voice. Later, Sara Ezekiel used a computer to master voice generation function, although this voice for communication was not at all like her own.
She was also able to continue her career as an artist, focusing on creating images on the computer. However, her two children have never heard their mother’s real voice since they were born.
In recent years, experts have been increasingly using technologies for creating digital versions of the original human voice. However, this technology requires long and high-quality recordings. However, even in this case, the voices will sound dull and monotonous.
According to Simon Poole, a representative of Smartbox, a British company that deals with communications in the medical field, in a comment to AFP, his company initially asked the woman for an audio recording lasting about an hour. People, who are expected, to lose their voice due to diseases, such as amyotrophic lateral sclerosis, experts recommend recording your own voice to preserve your identity and restore the ability to communicate fully.
However, before the advent of smartphones, the ability to use high-quality personal audio recordings was less common and more technically challenging. As a result Sarah Ezekiel managed to find only one, short and low-quality recording.
The clip from the 90s home video lasted only 8 seconds, was muffled, and was accompanied by background noise from the TV. This led Simon Poole to turn to a technology developed by ElevenLabs, an American company specializing in artificial intelligence. This technology not only recreates voice based on the simplest data, but also makes him sound more like a real person.
Simon Poole used one AI tool to highlight voice sample from the clip, and another tool trained on real voices to fill in gaps while creating the final version. The final result was as close as possible to Sarah Ezekiel’s real voice, including her London accent and a slight lisp.
“I sent her samples, and she wrote back that she almost cried when she heard them. She said, she played it to a friend, who had known her before she lost her voice, and it was like her own voice came back to her”, — says Simon Poole.
According to the British Association for Motor Neurone Disease, eight out of ten patients have voice problems after diagnosis. However, the pace, pitch, and tone of today’s computer-generated voices can be completely robotic.
“The real advantage of this new AI technology is that the voices have become really human and expressive, and they really bring back the humanity of the voice, that used to sound a little bit computer-like. In particular, if you develop a disease in adulthood and lose your voice, it is very important to be able to speak in your real voice, rather than using a prepared voice”, — Simon Poole is convinced.
Sources: AFP; TechXplore
Контент сайту призначений для осіб віком від 21 року. Переглядаючи матеріали, ви підтверджуєте свою відповідність віковим обмеженням.
Cуб'єкт у сфері онлайн-медіа; ідентифікатор медіа - R40-06029.