Emotion Recognition Through Speech

Speech is individuals’ primary contact form, and vocabulary is their primary tool. In social connections, emotion is essential. Given that we encounter interaction between humans and machines, it is crucial and difficult to identify the emotions in a speech. The way a person expresses their emotions changes depending on the feeling they are experiencing. When somebody expresses their emotions, each will have a distinctive intensity, pitch, and tone fluctuation that can be put together based on the matter. Therefore, a future objective of computer vision is the identification of spoken emotions. This paper aims to create an intelligent speech that recognizes emotions using a convolutional neural network. Multiple sections are utilized to identify feelings, and a classification algorithm is distinguished between happiness, sadness, anger, and amazement. The machine will transform the mathematical representation of human voice signals, go through its routine, and then show emotion. The data is a spoken specimen, and the Librosa library extracts the attributes from the voice sample. As a research dataset, we are employing the RAVDESS collection. This study demonstrates that all classifications for the data we used have an accuracy level of 68%.