they recreate their voice…digitally

2 million people a year are diagnosed with a speech disorder. Some of them use “My own voice”. Thanks to artificial intelligence, this French innovation reproduces the voice of those who have lost it.

It is a technology that will leave you speechless. Or rather, who recreates it. “My own voice” is a voice synthesis tool, invented by the Acapela group, which “makes” the digital voice of a voiceless person after a disease or an operation (amyotrophic lateral sclerosis, throat cancer, aphasia, cerebral palsy, dysarthria, laryngectomy… ). Available in more than twenty languages, including French, this service relies on “machine learning”. When the patient’s voice can still be heard, he can record fifty sentences on the platform. These will serve as the basis for the artificial intelligence to imitate its timbre, its accent, its intonations. He can then enter text directly on the “My Own Voice” page, which will be translated into an audio version with the pre-recorded voice signature.

2 million people with speech disorders

Each year, approximately 2 million people worldwide are diagnosed with a speech disorder, but only 2% are equipped with a communication aid, according to the Acapela group. In 2016, the creators of “My Own Voice” smelled a market. The winner of the digital health category at CES Las Vegas 2023, the Toulouse-based innovation now has almost 10,000 users worldwide, including 3,400 new customers in 2022 alone. emulators across the channel, success is still timid in France, confides Nicolas Mazars, its commercial director. However, the result is “impressive”, according to Olivier Goy, entrepreneur who suffers from Charcot disease (ALS). Affected by slurred speech, he was able to reproduce his voice from a simple podcast recorded before the first manifestations of his illness. “Long live technology in the service of health”, explains his digital voice revealed on the social network Linkedin. If most users adopt this device in anticipation, others because their voice is already turned off, “involve their relatives, a brother, a sister, with an approachable tone”, says Nicolas Mazars.

A fantastic result was fascinated by the promises of this jewel of technology and wanted to test it. It was our journalist, Clotilde Costil, who tried the experiment, the challenge was to be able to find her characteristic thin voice. It has been rendered almost identical to a barely noticeable robot veil. Even his relatives validated this copy “bluffing”. The only downside is that the customer’s digital journey still lacks fluidity. “This service does not require no special skills or equipment, apart from a computer, a headset and an internet connection,” says the Acapela group. It is true… in part. Admittedly, the tool is accessible from a material point of view, but it is not always intuitive. Our journalist, although familiar with new technologies, sometimes got lost during the various stages of the recording process, and only translated into French half the time.

A fourth version in 2023

Let future potential customers rest assured, these few hiccups should be history by the fourth version. This will see the light of day in 2023, according to Nicolas Mazars. On the other hand business model will remain the same: the initial stage to create your voice is free. “Only when you need it should you buy it,” the fire continues. So far, the device is not eligible for disability compensation (PCH) or reimbursed by health insurance, although “this is under way”, reassures Nicolas Mazars, without however giving a deadline. You therefore have to pay 99 euros per month or 999 euros per year (renewable) to get your personal digital voice, the one that (usually) never fades…

© The Acapela Group

“All rights of reproduction and representation reserved. © This article was written by Clotilde Costil, journalist”


Leave a Comment