Speech-based clock captures aging

October 27, 2026

Speech-based clock captures aging

At a Glance

  • Researchers found that many speech features such as pitch, timing, and word choice can help detect accelerated aging and dementia.
  • With further development, this approach could be a less expensive and easier way to measure how aging affects the body and brain.
Older couple sitting at a table celebrating great news.
The way a person talks could provide clues about the way their brain is aging, according to a new study.
SALMONNEGRO / Adobe Stock

Age is more than just how old someone is in years, or their “chronological” age. “Biological” age measures the effect aging has on the body, including the accumulation of cellular damage over the years. People whose biological age is older than their chronological age are aging more quickly than expected. This raises their risk for age-related health problems like dementia. 

Tools called aging clocks estimate the difference, or gap, between a person’s chronological and biological age. Aging clocks can use statistical models to predict biological age from behavior, imaging, or other data. The difference between predicted and chronological age reflects premature changes to a person’s body or brain. But they often use technologies that are expensive or not available in areas with few economic resources. 

An NIH-funded research team led by Drs. Adolfo García and Agustin Ibanez of the Global Brain Health Institute set out to learn whether a person’s speech could be used to estimate age gaps and their links to dementia. The study was published in Science Advances on September 30, 2026. 

The researchers analyzed speech recordings of 2,928 native Spanish speakers in five Latin American countries. About half of the participants had mild cognitive impairment, Alzheimer’s disease, or a form of dementia called frontotemporal dementia.  

The team developed a speech clock that used machine learning to detect patterns in participants’ speech. The models assessed hundreds of elements, such as speech pitch, timing, word properties, and emotional content. These and many other features related to sound and language were combined to predict a participant’s biological age and compute speech age gaps. 

The researchers found pitch and emotional content were most important for identifying people whose predicted age was older than their chronological age. Speech timing and the number of words helped identify people predicted to be younger than their age in years. 

The speech-based clock’s estimates of biological age were similar to those of other biological clocks that use brain scans or look at the pattern of certain chemical markers on DNA. This suggests the way someone speaks reflects how quickly their brain and body are aging. 

People with mild cognitive impairment or dementia were more likely to be biologically older as estimated by the speech-based clock. The age gap was larger for people with Alzheimer’s than those with mild cognitive impairment. The gap was even larger for people with frontotemporal dementia. The speech age gap was smallest among healthy participants. 

The team found that larger gaps between the two measures of aging reflected poorer performance on cognitive tests. And in people with Alzheimer’s, the larger the gap was, the more of an Alzheimer’s-related protein called p-tau217 they had in their blood. 

The researchers also assessed how a person’s social exposome affected biological age. The social exposome includes factors such as access to food and health care, education, and childhood experiences. They found that for healthy people and people with Alzheimer’s, the more adversity in someone’s social exposome, the higher their speech-predicted biological age was compared to their chronological age. 

The results suggest that speech-based clocks can detect patterns linked with accelerated aging. But long-term studies are needed to determine how well they can predict aging-related health problems like dementia.  

“If confirmed longitudinally and across populations, speech could ultimately become one of the most scalable tools for monitoring healthy and accelerated aging,” Ibanez says. 

—by Brandon Levy

Related Links

References

Speech clocks decode dementia phenotypes, social exposome, and biological aging. Hernandez H, Pedraza LK, Santamaria-Garcia H, Moguilner S, Legaz A, Prado P, Cuadros J, Amoruso L, Gonzalez L, Dellavale D, Espinoza-Puelles JP, Espinosa JP, Jarne C, Mattiussi F, Caccia M, Welford AS, Pelella N, Inchauspe J, Ferrante FJ, Pérez G, Maito MA, Rocatti G, Godoy ME, Migeot J, Orellana P, Caviedes A, Bruno M, Takada L, Slachevsky A, Behrens MI, Bruna B, Aguillon D, Zapata L, Avila-Funes JA, Custodio N, Miller B, Gorno-Tempini ML, Escudero SP, Reyes P, Hu K, de Oliveira MO, Coronel-Oliveros C, Cruzat J, Cardona JF, Corley M, Meier IB, Narayan VA, Tagliazucchi E, Baez S, Duran-Aniotz C, García AM, Ibanez A. Sci Adv. 2026 Oct 2;12(40):eaef9864. doi: 10.1126/sciadv.aef9864. Epub 2026 Sep 30. PMID: 42814823.

Funding

NIH’s National Institute on Aging (NIA), National Institute on Drug Abuse (NIDA), and Fogarty International Center (FIC); Davos Alzheimer’s Collaborative; Multi-Partner Consortium to Expand Dementia Research in Latin America; Alzheimer’s Association; Rainwater Charitable Foundation; Global Brain Health Institute; Agencia Nacional de Investigación y Desarrollo; JPI JPND-Care; DISCeRN 2025; FONDEF; Wellcome Trust; Spanish Ministry of Science and Innovation.