AI Voice Clock measures how quickly you’re aging using your voice

A “voice clock” developed by scientists can predict how much a person is aging based on the characteristics of their voice and the way they speak. The clock relies on hundreds of voice characteristics, including pitch and speaking speed, to estimate a person’s age.

An ‘inflammation clock’ can reveal the body’s biological age

Using the clock, the scientists calculated a “speech age gap” for individuals: the difference between a person’s age as predicted by the speech clock and their chronological age. Large age gaps in speech were strongly associated with cognitive problems in people, such as those that occur in dementia, suggesting that the new clock could be a useful tool for determining whether a person is aging faster than expected.

The results were published today in the journal Scientific advances¹.

“We can see enormous predictive value with a simple four-minute voice recording,” says neuroscientist Agustín Ibáñez of the Adolfo Ibáñez University of Santiago, co-author of the study.

The voice clock could be a boon for tracking the aging of people in low-resource areas because it does not rely on expensive or invasive technologies, such as brain scans and blood tests, says Jed Meltzer, a cognitive neuroscientist who specializes in language at the University of Toronto in Canada and was not involved in the study. “It’s very impressive work.”

Predicting age with speech

Aging clocks often use biological markers to gauge how quickly a person’s body is declining. For example, “brain clocks” use neuroimaging signatures to determine whether an individual’s brain is aging faster than their chronological age suggests. And “epigenetic clocks” look at patterns of methyl tags on a person’s DNA to estimate their biological age. But until now, researchers have not developed a speech-based clock. Such a tool could provide a useful window into the aging process, because speaking involves “a huge amount of brain work,” says Ibáñez.

How fast are you aging? What molecular “clocks” can tell you about your health

To create their clock, Ibáñez and his colleagues recorded 2,928 Spanish speakers from Argentina, Chile, Colombia, Mexico and Peru as they performed various vocal tasks. The group was a mix of healthy individuals and people with mild cognitive impairment, Alzheimer’s disease or other forms of dementia. Researchers used machine learning algorithms to extract from audio recordings more than 700 speech characteristics that can change with aging and dementia, such as pitch and vocabulary range. They then used this data to train their vocal clock model to predict the age of each study participant.

Overall, the vocal clock could distinguish between healthy individuals and those with some form of cognitive impairment. The clock, for example, categorized the speech of people with cognitive problems as older than would be expected given their chronological age. Healthy people were generally assessed as having speech appropriate to their chronological age.