Determination and evaluation pitch harmonics parameters with emotions classification

People´s coding and decoding of nonlinguistic speech attributes is usually subconscious with an enhanced psychological element and it brings spontaneousness, more information and beauty of speech act into the speech communication. Therefore, these attributes are extremely important within the human-machine speech communication. Emotions are definitely one of the most important nonlinguistic speech attributes. Each speaker´s emotion forms vocal chords and vocal system that is acoustically shown through variations of the speech signal parameters. In this paper, the glottal speech during the pronouncing vowel /a/ are analyzed. This is done by determination and evaluation of structure of pitch harmonics parameters for three emotional states: neutral, anger and speech under stress. The difference between structure pitch harmonics parameters of glottal speech is very informative about speaker´s emotional state. Implemented pitch harmonics normed indexes articulately and easily describe harmonics structure of glottal speech and they can be significant element in a classification of emotions.