Voice Age model estimates chronological age through speech analysis

Researchers have developed a non-invasive method to estimate chronological age through vocal analysis. According to a study published in the journal npj Aging, a machine learning model named ‘Voice Age’ can extract significant information about a person's age from just 30 seconds of recorded speech.

The study involved 6,979 Hebrew-speaking adults between the ages of 40 and 70. The model demonstrated high accuracy, estimating age with an average deviation of approximately four years for women and 4.4 years for men. These vocal changes occur because speech relies on the coordinated function of the respiratory system, larynx, muscles, and nervous system, all of which undergo physiological changes during aging.

This research builds on previous findings suggesting that speech patterns can indicate various aspects of physical functioning, such as mobility, balance, and fatigue. While subtle changes in articulation and speech complexity may be imperceptible to the human ear, computational models can identify these patterns to serve as potential biomarkers for aging.