06/10/2026
"You answer the phone and hear a single word—“Hello”—and instantly an impression forms. Without thinking, you know whether the voice is familiar or belongs to a stranger. You also make guesses about the speaker’s gender, age, or current mood. For most listeners, this recognition happens rapidly and automatically.
Voice carries information beyond the literal meaning of words. Embedded in the acoustic signal are cues that listeners have learned to decode with remarkable speed. We extract not only what is being said but who is saying it.
What goes up but never down?
Age.
Age presents a particularly interesting riddle. It is something we all share—a continuous process unfolding across the entire lifespan. We are surrounded by aging speech, including our own, our family’s, our friends’, and the speech of the media we watch and listen to. Every voice we hear is in the process of aging. Given this constant exposure, we might expect people to be expert judges of vocal age, finely attuned to the acoustic markers that distinguish a voice in its twenties from one in its sixties.
Early research tested this expectation by playing recordings of many different speakers, ranging in age from children to older adults, to groups of listeners, who were then asked to estimate each speaker’s age. Overall, listeners were quite adept: their estimates lined up strongly with actual age (r = 0.85). But the errors followed a consistent pattern. Listeners systematically guessed that younger speakers were older than they actually were and that older speakers were younger than they actually were (Hunter et al., 2016).
This pattern of overestimation and underestimation converged at around age 50. Guesses were most accurate for speakers between ages 35 and 55, with about 6 years of error. However, errors for the oldest speakers (ages 85–90) exceeded 11 years (Hunter et al., 2016). This pattern appeared across multiple independent studies, suggesting that it reflects something fundamental about how we perceive vocal age.
The systematic nature of these errors hinted at something interesting: perhaps listeners were not tracking age continuously but were sorting voices into broad age categories. But these studies, by averaging across many different speakers, could not reveal how individual voices age.
Does a single person’s voice show gradual, detectable changes year by year? Or does something more complex happen? And what information are listeners using to judge age?
Answering these questions required a different approach, one that followed the same voice across decades."
from the intro to "How we hear age in the Human Voice" by Mark L. Berardi, Sarah Hargus Ferguson, Eric J. Hunter, and Benjamin V. Tucker. First published in Acoustics Today, Spring 2026
