Abstrakt
This study concerns the influence of basic prosodic parameters on the perception and identification of emotions. Emotional sentences have been generated by the TTS system with high naturalness of speech using modification by hand of pitch contour, intensity and duration as well as their combinations.
The prosody of sentences has been modified in order to express four emotions: anger, fear, joy and boredom, and this for both male and female voices. Subsequently, the sentences with emotions modelled by means of prosody have been applied to the listening tests, which uncovered the importance of different parameters for the identification of different emotion.
The results show that the identification of different emotions is based on relevant changes of different parameters and their combinations.