Research on Speech Emotion Recognition Analysis Based on Deep Learning

Abstract

Speech is a common way of communication in human daily life, which is very natural and easy to understand. In the process of communication, one will also express their emotions to the other party. Through emotional expression, the distance between people is smaller and communication is more convenient. Speech emotion recognition is also the key to human-computer emotional interaction, and effective recognition of speech emotions can improve speech intelligibility. The number of speech datasets is usually large, with several G or even larger. This article proposes a method of preprocessing the speech dataset to extract feature data folders, and then using the feature data folder directory as input to train the model. This method helps to improve the efficiency, repeatability, and manageability of data processing.