Keywords:emotion, voice, facial expression
We propose a method to detect insincere utterances based on facial expression and tone of voice. This paper develops our previous method by changing features. In order to decrease the bad influence due to the different mouth-movement at free sentence task, the proposed method removes the features of coordinates for mouth. Furthermore, the type of acoustic feature are significantly increased from 3 to 1,027. The experimental result represented that the recall of free sentence task was improved 15 point over the previous method. Furthermore, the accuracy of fixed sentence task was improved 5 point over the previous method.
Authentication for paper PDF access
A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.