A comparative study of silence and non silence regions of speech signal using prosody features

摘要

The objective of the present work is to develop an emotion based speech recognition system with the knowledge of speech statistics and prosodic features. A detailed study over silence and non silence regions of a speech signal helps in increasing the performance accuracy of the system. The idea of removing the silence regions from the speech information with a predefined threshold is taken forward in this paper. Prosody features deals with the auditory qualities of the sound and it can also reflect aspects of meaning, intention and emotional state of the speaker. Emotion depicting features like energy, pitch and duration are considered. These features are classified using a conventional GMM. The performance of the present system is evaluated using Berlin emotion speech corpus. The percentage-accuracy of the system has increased from 21 to 36%.