16 Seconds of Broadcast Audio Brings Back ALS Patient's Voice

Just 16 seconds of audio preserved in a broadcast from 14 years ago has brought back the voice of a patient who lost his speech to illness. Paul Francois, 67, of Vermont, who suffers from ALS (a neurological disease that gradually robs people of muscle control and the ability to speak and swallow), has regained his voice thanks to technology from AI voice cloning company ElevenLabs — with the source material being a short audio clip extracted from a local broadcast video recorded 14 years ago.

Francois was diagnosed with ALS in 2019. As his muscles and motor functions weakened, his ability to speak steadily faded. His wife Kris took on the role of caregiver, with their son Ben lending support. Communication relied on eye-tracking technology: software would read aloud letters one by one as he selected them with his eye movements. The problem was the sound itself. Kris described the existing synthesized voice as "very monotone and robotic."

Vermont has the highest rate of ALS per capita in the United States, with roughly 27 people in the state living with the disease.

ElevenLabs offers this technology free of charge to ALS patients. The company handles voice cloning at no cost, while the nonprofit organization Bridging Voices connects families with the company.

The case of patients who once had to rely on robotic synthesized voices regaining their natural, original voices demonstrates how AI voice technology can be applied to real life.