Speech clock uses machine learning to measure biological aging

Researchers have developed a machine learning-based “speech clock” that can estimate biological aging by analyzing linguistic and acoustic patterns. A study published in Science Advances utilized data from nearly 3,000 Spanish-speaking adults across Argentina, Chile, Colombia, Mexico, and Peru to track how speech characteristics correlate with brain, cognitive, and social aging.

The system analyzes hundreds of parameters, including speech speed, pauses, voice inflection, emotional content, and word choice. By comparing a person’s chronological age to the age estimated from their voice, researchers can identify a “vocal gap.” A significant gap—where a voice sounds more “mature” than the individual's actual age—has been linked to accelerated brain aging and signs of cognitive decline, such as Alzheimer’s or frontotemporal dementia.

This technology offers a potential low-cost, non-invasive, and automated biomarker for aging. Unlike epigenetic clocks or neuroimaging, which often require expensive equipment or specialized settings, speech analysis could provide an equitable tool for large-scale aging research and dementia monitoring, particularly in low-resource or underrepresented regions.