The component analyses and properly reacts to laughter allowing an ECA to join his conversational partners’ laughters, offering the ability to react to surrounding acoustic events different from speech.
Appropriate reaction is supported by an analysis phase that supports the choice of suitable answers that a conversational agent can provide the user, selected among similar laughters according to patterns similar to the input laughter.
Availability of the component: Reference contact is Jerome Urbain of Universite' de Mons.
Notes: MediaCycle, at the core of acoustic awareness component, is still under development. It was first developed under Mac OS and is currently moved to Windows. Both the library of features and the similarity engine can be provided as executables or compiled libraries for non-commercial use upon request.
The component is used in a Proof-of-concept application: the AVLaughter Machine, in conjunction with Smart Sensor Integration component and Emotional Attentive ECA. It supports ECA to be driven with the selected utterance. A MediaCycle application was experimented, accepting queries by audio signals: during the CALLAS training session at eNTERFACE 2009 an audiovisual laughter database has been recorded with acoustic but facial movements of people laughing. More information is in the following Papers and Articles:
Acoustic Awareness component
European Commission • Sixth Framework Programme - CALLAS Project © 2006. For more info, please read Job Posts Milan
