This repository contains the source code of our project that implements Machine Learning techniques and algorithms for a Speech Emotion Recognition Project. The .wav file of a speech segment is analyzed and is classified by its emotion.
A multi-class classification problem was implemented with 5 classes:
- fear
- anger
- sadness
- happiness
- disgust
Our main application is implemented in the Jupyter file named SER_ML.ipynb. We have added comments in each cell and before the definition of complex functions.
If you want to run the notebook follow these steps:
- Open it in Colab using the embedded link
- Use the data as we have modified them. You can find them at this link. You could create a shortcut of this location in your drive.
- Create a copy of the notebook in your drive and run!
We recommend you use a runtime with GPU so ThunderSVM library can be used and accelarate the SVM computations.