Integrated Approach for Emotion Detection via Speech and Text Analysis

Abstract

This paper aims to provide a comprehensive solution for effective reviews using deep learning models. Customers often have difficulty to find accurate reviews of the things they are interested in. The proposed framework implements a review mechanism to address this problem, which will give customers relevant reviews based on video reviews supplied in the product description. The goal of this system is to turn video reviews into a particular rating so that viewers may get a summary of the review without having to watch the full thing by simply glancing at the rating. Deep learning neural networks are used by the model for both text and audio processing in order to achieve this. The well-known RAVDESS dataset serves as the basis for the audio model's training and offers a wide range of emotional expressions. The suggested system uses two methods to provide reviews: text-based natural language processing and audio frequency spectrograms. By utilizing these two techniques, it may provide consumers accurate and trustworthy ratings while guaranteeing that the review procedure is not impeded. The aim is achieved with high accuracy to ensure that users can make informed decisions when purchasing products based on the provided reviews. With the aid of this review system, customers will be able to quickly find out crucial details about a product they are interested in, thus increasing their pleasure and loyalty.