Research demonstrates how large language models can accurately assess speech recall across languages, addressing critical gaps in speech comprehension testing.

• Enables language-agnostic assessment of speech recall using naturalistic, story-like content
• Overcomes limitations of traditional speech tests that disadvantage non-native speakers
• Provides automated, time-efficient scoring with high accuracy compared to human evaluation
• Shows particular promise for evaluating speech comprehension in older adults

This innovation offers significant medical value by providing more accessible, accurate diagnostic tools for speech comprehension difficulties, which are especially common among aging populations.

Language-agnostic, automated assessment of listeners' speech recall using large language models