This repository contains code and resources for performing Speech Emotion Recognition (SER) using neural networks and Long Short-Term Memory (LSTM) models. The RAVEDNESS dataset is utilized for training and evaluation.
- RAVEDNESS Dataset: Link to the dataset
Please download the dataset and place it in the appropriate directory before running the code.
We have implemented two main models for SER:
-
Neural Network (NN) Model: A feedforward neural network designed for SER.
-
LSTM Model: A Long Short-Term Memory (LSTM) model tailored for sequence data in SER.
You can find the code for these models in their respective directories.
Follow these steps to run the code:
-
Install the required dependencies by running:
pip install -r requirements.txt -
Organize the dataset as specified in the dataset section.
-
Train and test the models by running the respective scripts.
-
Evaluate the model performance, and visualize the results.
We have obtained the following results:
- Confusion Matrix Neural Network:
- Confusion Matrix LSTM :
- (
