Speech Emotion Recognition Using Deep Learning Techniques
Mr. Ajay Kumar Bansal1, Ms. Shivangi2, Asgar Aizaz Peerzada3
Author Affiliations
Dept. of Computer Applications, Lovely Professional University, Phagwara, Punjab, India
Abstract: With the increase in the enhancement of materialistic things in the world has turned the way of living of people extremely. Due to which there is tremendous increase in stress in day-to-day life. Such immense effect on the emotional state of people, has led to increase the importance of sentimental analysis. If emotional state of the person is not neutral, then one is unable to make any sort of decision related to either of profession or life. Emotional state of a person has direct relation with respect to any sort of action a person is going take in his living. So, this creates a sentimental analysis an essential activity that needed to be performed, whether it comes to the context of student siting in the class in order learn something new or any employee who is sitting in the office and needs to take important business decision in profession. In context with this Speech Emotion Recognition (SER) has made things easier to first analyze the sentimental state of a person and then try to turn one into neutral state to make one’s living efficient. The proposed model is first going to check the current emotional state with the help of speech which carries various information like tone, pitch and speaking rate. Deep Learning technique named as LSTM (Long-Term Short-Term Memory) which is a part of RNN (Recurrent Neural Network) that is able to handle long-term dependencies in sequential data and application of sentimental analysis and speech emotion recognition, which was found out be most effective in such analysis through this paper.
Keywords: Speech emotion Recognition (SER), LSTM, RNN
