Speech Emotion Recognition with Web Speech API
M. Vijay Anand, A Rajasekaran, Rajendra Thilahar C, N. Bharathiraja · 2023
Speech is one of the natural ways that we as humans communicate with one another. We rely on it so much that we have come to understand its significance in electronic communication, including text messages and emails, where we frequently utilize emojis to convey the emotions we are feeling at the time. Since emotions are a major component of communication, it is critical to identify and analyze them in the digital world of remote communication that we live in today. We use the SER technique because emotion detection is a difficult task because emotions are subjective. The goal of Speech Emotion Recognition (SER) is to identify the affective components of speech, independent of its semantic content. The goal of research on automatic emotion recognition systems is to develop effective, real-time techniques for identifying emotions in a range of application domains, such as caller-agent conversation analysis and interactive voice assistants. In this project, Web-Speech API—pre-existing software for analyzingthe nuances and annotations of speech—is used to analyze and identify the emotional aspects of speech.Further we will be recognizing the words we speak using the scores which are predefined in the AFINN database. The outputwill be displayed in the webpage in the form of emojis as we speak in real-time. The approximate accuracy obtained in this project is around 80%-85% .Thus the sentiment in any content published as a speech or a live recording can be analysed through this project.