No abstract

No abstract

Building on the success of the ADReSS Challenge at Interspeech 2020, which attracted the participation of 34 teams from across the world, the ADReSSo Challenge targets three difficult automatic prediction problems of societal and medical relevance, namely: detection of Alzheimer's Dementia, inference of cognitive testing scores, and prediction of cognitive decline. This paper presents these prediction tasks in detail, describes the datasets used, and reports the results of the baseline classification and regression models we developed for each task. A combination of acoustic and linguistic features extracted directly from audio recordings, without human intervention, yielded a baseline accuracy of 78.87% for the AD classification task, an MMSE prediction root mean squared (RMSE) error of 5.28, and 68.75% accuracy for the cognitive decline prediction task.

No abstract

The results of a comparison between three different speech types-On-Talk, speaking to a computer, Off-Talk Self , speaking to oneself and Off-Talk Other, speaking to another person-uttered by subjects in a collaborative interlingual task mediated by an automatic speech-to-speech translation system, are reported here. The characteristics of the three speech types show significant differences in terms of speech rate (F2,2719 = 101.7; p < 2e − 16), and for this reason a detection method was implemented to see if they could also be detected with good accuracy based on their acoustic and biological characteristics. Acoustic and biological measures provide good results in distinguish between On-Talk and Off-Talk, but have difficulty distinguishing the sub-criteria of Off-Talk: Self and Other.

Building on the success of the ADReSS Challenge at Interspeech 2020, which attracted the participation of 34 teams from across the world, the ADReSSo Challenge targets three difficult automatic prediction problems of societal and medical relevance, namely: detection of Alzheimer's Dementia, inference of cognitive testing scores, and prediction of cognitive decline. This paper presents these prediction tasks in detail, describes the datasets used, and reports the results of the baseline classification and regression models we developed for each task. A combination of acoustic and linguistic features extracted directly from audio recordings, without human intervention, yielded a baseline accuracy of 78.87% for the AD classification task, an MMSE prediction root mean squared (RMSE) error of 5.28, and 68.75% accuracy for the cognitive decline prediction task.

No abstract

No abstract