Fig 1.
System framework of role-playing and feedback through interaction with computer avatar.
Fig 2.
Extracted facial landmark points using face tracker.
Fig 3.
Audiovisual feedback provided by automated social skills trainer: User video, overall score, comparison with models, and positive comments.
Fig 4.
Improvement of overall narrative scores in audio and audiovisual groups.
Error bars indicate standard error (*: p<.05).
Fig 5.
Overall narrative scores of pre- and post-training.
Participants are indicated by color. We added a small amount of noise to separate identical points.
Table 1.
Relationship between participant ages and scores.
Table 2.
Correlation coefficient to overall narrative skills.