Laughs and emotions

Juan José Moreira Pérez | Researcher

In recent weeks, Gradiant has appeared in different media for being the architects of the happiness meter that can be accessed in, where users can upload their photos and the system determines the level of happiness of people present in it from their smiles. This is an example of application of affective computing technologies in which Gradiant is working. This field comprises research and development of tools to determine the mood of people through audiovisual information. Within this field, another line of work in our company is the detection of laughter by means of acoustic information.

Not so long ago we published a technological report on recognition of emotions in the voice. One of the clearest indicators of positive moods is the presence of laughter in the utterances.

Given the strong correlation between the presence or absence of laughter and mood, the interest in the development of mechanisms that serve to detect this feature of human communication has been increasing within the field of affective computing.

Detection of laughter is a complex problem as there are many different types of laughter, from loud laughter to almost inaudible chuckles. Some are voiced while others appear as a simple periodic exhalation of air through the nose.

The attendants to the Summer School AERFAI Summer School on Pattern Recognition in Multimodal Human Interaction, organized by Gradiant, the University of Vigo and University of Oporto, which was held at Vigo from June 4 to June 8, could attend the lecture of Björn W. Schuller on speaker state and trait analysis. Within a more general section on detection of non-linguistic vocalizations, Björn explained the techniques used to detect laughter and showed some of the obtained results.

The detection of laughter has multiple applications. One is to control and monitor the reactions of the audience to a particular show or multimedia content to determine whether they like it or not, or even modify the content to be displayed depending on the perceived mood.

The first commercial applications of this style may not take too long to hit the market. Recently Sony has patented the detection of laughs for their videogame consolesusing visual and acoustic information obtained from a webcam and a microphone to identify the user and try to determine his/her reaction to the content being played at that time.

Laugh detection and audio processing open the door to multiple fields of application, not only in the entertainment field but also in healthcare and Ambient Assisted Living (AAL), where affective computing technologies are being progressively adopted.