Speech Emotion Recognition Using Recurrent Neural Network

Abstract

Speech Emotion Recognition is a project that identifies the emotion of a person based on his/her voice. This project is based on the Recurrent Neural Network(RNN), which uses different modules for emotion recognition. The classifiers are used to identify various emotions namely happiness, anger, sadness, disgust, fear, neutral state and surprise.

In this there will be a recorded voice of a person and our system identifies the emotion from that recorded audio. Various features are extracted from the voice using the LIBROSA package of language python. Our system uses the same phenomenon that animals like horses and dogs use to understand human emotion. It analyzes the audio files in WAV format and returns the intended outcome i.e. the emotion.