Publication Details

Abstract

The key issue of emotion detection is choosing the speech database, identification of various variables connected to speech, and model selection. Emotional speech recognition has advanced from a routine activity to a crucial part of Human-Computer Interaction (HCI). Mel Frequency Central Coefficient, or MFCC, is employed in this article to extract features. The approach is based on recurrent neural networks (RNN) and long short-term memories (LSTM). The database is TESS (Toronto Emotional Speech Set). There are 7 emotions in the TESS dataset. They are indifferent, fearful, happy, surprised pleasantly, sad, and angry. This essay makes use of these 7 emotions. By utilizing this model, an accuracy of about 83% is obtained.