F0 feature analysis of communication between elderly individuals for health assessment

Abstract

This study explores a system that estimates the health condition of an elderly individual using nonverbal information from daily conversations. We have already confirmed the effectiveness of using the fundamental frequency (F0) to estimate the atmosphere of a conversation between young individuals. A smooth conversation has a tendency for its average value of F0 (Ave-F0) to increase slightly, and its standard deviation value (SD-F0) to increase significantly compared with a non-smooth conversation. The differences are significant when using a t-test, where the confidence level is 95%. We confirmed that Ave-F0 and SD-F0 are useful in separating laughter utterances from usual speech utterances. In this paper, we report on the acoustic analysis results of a free conversation between elderly individuals, and compare it with the analysis results of young individuals. We describe the possibility of estimating a health condition using F0 characteristics.