Speech Emotion Recognition Using MLP Classifier

Project Overview: This project aims to build a speech emotion recognition system using the RAVDESS dataset. The system extracts various audio features (MFCC, Chroma, Mel Spectrogram) from audio files and classifies emotions using a Multi-Layer Perceptron (MLP) classifier. The model identifies emotions such as calm, happy, fearful, and disgust.

Dataset: The RAVDESS dataset is used in this project, containing audio recordings of emotional speech with the following emotions:

Neutral, Calm, Happy, Sad, Angry, Fearful, Disgust and Surprised

For this project, only the emotions calm, happy, fearful, and disgust are used for classification.

Features Extracted: MFCC (Mel Frequency Cepstral Coefficients), Chroma (Chroma Short-Time Fourier Transform) and Mel Spectrogram

These features capture various characteristics of the audio signals, enabling the classification of emotions.