Abstract

Emotion plays a significant role in communication. This work deals with an analysis of the sets of Kannada emotional sentences uttered by two speakers (one Male and one female). The data set used in this analysis is from KES (Kannada Emotional Speech) database, which consists of neutral, happy, sadness, anger and fear emotional sentences. The emotional analysis of acoustic features on KES database is carried out to build an effective emotion recognition and emotion conversion system. The features used of analysis of emotional speech are pitch, duriation, intensity, formats and spectral energy at sentence level. Results shown that prosodic and spectral parameters play significant role in expression of emotions. The acoustic values are in correspondence with the general knowledge for male and female voices. The analysis result obtained will be used for emotional speech conversion.