Published Paper Details:

In recent years, emotion recognition has become a rapidly increasing research subject. Machines, unlike humans, lack the ability to perceive and express emotions. However, automatic emotion recognition can improve human-computer connection, minimizing the need for human intervention. The manufacturing properties of these signals are extracted using signal processing methods. The studies use the source's instantaneous fundamental frequency (F0), the system's formants and dominant frequencies, the zero-crossing rate (ZCR), and the combined features signal energy. Zero-frequency filtering (ZFF) is used to produce F0, while LP spectrum is used to obtain formants and dominating frequencies. A rectangular window of 200 samples is used to derive short-time signal energy (STE) and ZCR in the voiced and unvoiced regions. The main goal of SER is to improve the human-machine interface. It can also be utilised in lie detectors to check a person's psychophysiological condition. Speech emotion recognition has recently found applications in health and forensics. In this work, pitch and prosody traits are used to distinguish seven emotions. The majority of the speech features used in this study are time-domain features. The emotions were classified using a Long Short Time Memory (LSTM) classifier.