Recognition of human emotion using computer has become a notable focus in research area. Specifically, emotion recognition through human speech is important, as voice is the foremost communication tool of human beings. This paper emphasizes different techniques that focus mainly on detecting emotional states through vocals. Besides, outlook for feature extraction methods from existing speech data sets and ml methods with a special prominence on classifiers are hypothesised. Various types of classifiers were used to evaluate previously constructed speech emotion recognition systems in this study. Classifiers that differentiate between emotions such as surprise, rage, happiness, sadness, fear, disgust, and so on. A rigid model is built using Tensar flow, Keras. By varying the datasets and physical audio from human beings, we tend to maximize the accuracy. This model will be helpful for officials in the sector of security.
Software And Hardware