PCA and ICA Fusion Applied in Speech Emotion Recognition (SER)

This repository demonstrates the application of Principal Component Analysis (PCA) and Independent Component Analysis (ICA) in Speech Emotion Recognition (SER).

Datasets

To test these scripts, you will need to download the following datasets:

1. SAVEE - Can be downloaded in https://www.kaggle.com/datasets/ejlok1/surrey-audiovisual-expressed-emotion-savee
2. RAVDESS Can be downloaded in https://www.kaggle.com/datasets/uwrfkaggler/ravdess-emotional-speech-audio
3. Berlin EmoDB - Can be downloaded using audb.

How to Use

1. Generate Files

Run the following scripts to create .xlsx and .csv files:

• <database_name>_pca_ica_components.py
• <database_name>_pca_ica_components_kw.py

Replace <database_name> with the name of the dataset you are using see name of the files.

2. Generate Graphs

After generating the necessary files, you can create graphs by running the plot.py scripts.

Example

To create PCA and ICA components for the SAVEE dataset:

python SAVEE_pca_ica_components.py
python SAVEE_pca_ica_components_kw.py

To plot the results:

python plot_emodb.py
python plot_savee.py
python ravdess_graph_plot.py

Notes

• Ensure that the required datasets are downloaded and properly set up before running the scripts.
• The scripts assume that the datasets are organized in their respective directories.

Acknowledgements

• SAVEE - Surrey Audio-Visual Expressed Emotion Database
• RAVDESS - Ryerson Audio-Visual Database of Emotional Speech and Song
• Berlin EmoDB - Berlin Database of Emotional Speech

Feel free to reach out if you encounter any issues or have questions.

---

---