Estimating Age-Dependent Degradation Using Nonverbal Feature Analysis of Daily Conversation

In this paper, we study a system that estimates the degree of decline in the driving ability of elderly people using non-verbal information from daily conversations. It is necessary for us to find the factors that would affect the calculation of the degree of decline that has reached a problematic level for functioning daily life. We focus on the cases where elderly people cannot understand their partner’s speech as their hearing and concentration abilities decrease with age. We analyze the relationship between the degree of understanding of the partner’s speech and the non-verbal characteristic of the response scene. Based on the results of the acoustic analysis of each utterance, the fundamental frequency (F0) and acoustic power levels of when a person can understand their partner’s speech tend to be higher than those when they cannot. The analysis of the synchronization of the head motions shows that brightness value of difference image when a person can understand their partner’s speech is also higher than when they cannot. These results indicate that these non-verbal factors are effective in estimating the decline in the hearing and concentration abilities of the elderly.