Laughter Inquisition in Affect Recognition

Laughter and humor are major ingredients of humanity but does not have any black and white authenticate. Laughter is a physiological process, which activates facial, respiratory and laryngeal muscles. Laughter may occur instinctively in response to humor or to appropriate emotional or sociological stimuli and can also be elicited upon command – voluntary, contrived or faked laughter. Exploring the pattern of laughter is an intricate and arduous errand. Speech and laughter are quite disparate in the vocalization, duration and in regularity of its occurrence. This work is a convincing attempt to catalog the different types of laughter as positive laughter, negative laughter, laughed speech and vague category. Feature selection was done using Radial Basis Boltzmann Machine Network, categorization was done using an ErgodicHidden Markov Model. The corpus used for experimental study was the International Computer Science Institute (ICSI) Meeting Corpus from which 30 meetings were taken as the data set. The technique used has yielded a persuasive result in categorizing the types of laughter.